Re: [dpdk-dev] drops while transmitting to the kni using rte_kni_tx_burst()

[Shirley Avishour]

Hi Jason,
You are right but I've tried to work in multiple mode as well and it did
not change the result.
I hope this issue will be solved.

Thanks,
Shirley.

On Fri, Jan 20, 2017 at 9:48 PM, Jason Kwon  wrote:

> Sorry to hijack the thread, but I was wondering, is the binding of the KNI
> kernel thread only applicable in multiple kthread mode?  Does it apply in
> single kernel thread mode?  Looking at the 16.07.2 source, I saw the
> following:
>
> /**
>  * Create a new kernel thread for multiple mode, set its core
> affinity,
>  * and finally wake it up.
>  */
> if (multiple_kthread_on) {
> kni->pthread = kthread_create(kni_thread_multiple,
>   (void *)kni,
>   "kni_%s", kni->name);
> if (IS_ERR(kni->pthread)) {
> kni_dev_remove(kni);
> return -ECANCELED;
> }
> if (dev_info.force_bind)
> kthread_bind(kni->pthread, kni->core_id);
> wake_up_process(kni->pthread);
> }
>
> Thanks,
>
> Jason
>
> On Mon, Jan 16, 2017 at 6:42 AM, Ferruh Yigit 
> wrote:
>
>> On 1/16/2017 12:20 PM, Shirley Avishour wrote:
>> > Hi,
>> > I have an application over dpdk which is consisted of the following
>> threads
>> > each running on a separate core:
>> > 1) rx thread which listens on in a poll mode for traffic
>> > 2) 2 packet processing threads (for load balancing)
>> > 3) kni thread (which also runs on a separate core).
>>
>> This is kernel thread, right? Is it bind to any specific core?
>> Is it possible that this thread shares the core with 2nd processing
>> thread when enabled?
>>
>> >
>> > the rx thread receives packets and clones them and transmit a copy to
>> the
>> > kni and the other packet is sent to the packet processing unit (hashing
>> > over 2 threads).
>> > the receive traffic rate is 100Mbps.
>> > When working with single packet processing thread I am able to get all
>> the
>> > 100Mbps towards the kni with no drops.
>> > but when I activate my application with 2 packet processing threads I
>> start
>> > facing drops towards the kni.
>> > the way I see it the only difference now is that I have another threads
>> > which handles an mbuf and frees it once processing is completed.
>> > Can anyone assist with this case please?
>> >
>> > Thanks!
>> >
>>
>>
>

Re: [dpdk-dev] [PATCH v4] doc: add known uio_pci_generic issue for i40e

[Mcnamara, John]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Jeff Guo
> Sent: Sunday, January 22, 2017 2:04 AM
> To: Zhang, Helin ; Wu, Jingjing
> 
> Cc: dev@dpdk.org; Guo, Jia 
> Subject: [dpdk-dev] [PATCH v4] doc: add known uio_pci_generic issue for
> i40e
> 
> When bind the ``uio_pci_generic`` module in X710/XL710/XXV710, the result
> is failed. The ``uio_pci_generic`` module is not supported by
> X710/XL710/XXV710.
> 
> Signed-off-by: Jeff Guo 


Acked-by: John McNamara 

Re: [dpdk-dev] [PATCH 2/3] vhost: fix long stall of vhost-user negotiation

[Maxime Coquelin]

On 01/22/2017 09:46 AM, Yuanhan Liu wrote:

Setting up the mapping from GPA (guest physical address) to HPA (guest
physical address) could be very time consuming when the guest memory is
backened with small pages (4K). The bigger the guest memory, the longer
it takes. This could lead a very long vhost-user negotiation.

Since the mapping is only needed in zero copy mode so far, we could
avoid such time consuming settup when zero copy is turned off (which is
the default case).

It's actually a workaround, a right fix might be to start a new thread,
and hide the big latency there.

Fixes: e246896178e6 ("vhost: get guest/host physical address mappings")

Cc: sta...@dpdk.org
Signed-off-by: Yuanhan Liu 
---
 lib/librte_vhost/vhost_user.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)


Reviewed-by: Maxime Coquelin 

Thanks,
Maxime

Re: [dpdk-dev] [PATCH v2] ethdev: fix MAC address replay

[Igor Ryzhov]

Hello Steve,

Thank you for all the fixes.
But I think I noticed another one issue in MAC replay process.

Pool number is extracted only once:

if (RTE_ETH_DEV_SRIOV(dev).active)
pool = RTE_ETH_DEV_SRIOV(dev).def_vmdq_idx;

But when MAC address is added using rte_eth_dev_mac_addr_add several
different pool numbers can be used.
Shouldn't we extract pool number for each MAC address separately from
mac_pool_sel array during restoration process?

Best regards,
Igor

On Sat, Jan 21, 2017 at 1:23 AM, Steve Shin  wrote:

> This patch fixes a bug in replaying MAC address to the hardware
> in rte_eth_dev_config_restore() routine. Added default MAC replay as well.
>
> Fixes: 4bdefaade6d1 ("ethdev: VMDQ enhancements")
>
> ---
> v2: Added default MAC replay & Code optimization
>
> Signed-off-by: Steve Shin 
> ---
>  lib/librte_ether/rte_ethdev.c | 32 
>  1 file changed, 16 insertions(+), 16 deletions(-)
>
> diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
> index 4790faf..150f350 100644
> --- a/lib/librte_ether/rte_ethdev.c
> +++ b/lib/librte_ether/rte_ethdev.c
> @@ -931,7 +931,7 @@ rte_eth_dev_config_restore(uint8_t port_id)
>  {
> struct rte_eth_dev *dev;
> struct rte_eth_dev_info dev_info;
> -   struct ether_addr addr;
> +   struct ether_addr *addr;
> uint16_t i;
> uint32_t pool = 0;
>
> @@ -942,23 +942,23 @@ rte_eth_dev_config_restore(uint8_t port_id)
> if (RTE_ETH_DEV_SRIOV(dev).active)
> pool = RTE_ETH_DEV_SRIOV(dev).def_vmdq_idx;
>
> -   /* replay MAC address configuration */
> -   for (i = 0; i < dev_info.max_mac_addrs; i++) {
> -   addr = dev->data->mac_addrs[i];
> +   /* replay MAC address configuration including default MAC */
> +   if (*dev->dev_ops->mac_addr_set != NULL) {
> +   addr = &dev->data->mac_addrs[0];
> +   (*dev->dev_ops->mac_addr_set)(dev, addr);
> +   }
>
> -   /* skip zero address */
> -   if (is_zero_ether_addr(&addr))
> -   continue;
> +   if (*dev->dev_ops->mac_addr_add != NULL) {
> +   for (i = 1; i < dev_info.max_mac_addrs; i++) {
> +   addr = &dev->data->mac_addrs[i];
>
> -   /* add address to the hardware */
> -   if  (*dev->dev_ops->mac_addr_add &&
> -   (dev->data->mac_pool_sel[i] & (1ULL << pool)))
> -   (*dev->dev_ops->mac_addr_add)(dev, &addr, i,
> pool);
> -   else {
> -   RTE_PMD_DEBUG_TRACE("port %d: MAC address array
> not supported\n",
> -   port_id);
> -   /* exit the loop but not return an error */
> -   break;
> +   /* skip zero address */
> +   if (is_zero_ether_addr(addr))
> +   continue;
> +
> +   /* add address to the hardware */
> +   if (dev->data->mac_pool_sel[i] & (1ULL << pool))
> +   (*dev->dev_ops->mac_addr_add)(dev, addr,
> i, pool);
> }
> }
>
> --
> 2.9.3
>
>

[dpdk-dev] [PATCH] mk: remove default toolchain prefix for ThunderX

[Thomas Monjalon]

The environment variable CROSS must be set when using a cross-toolchain.
However it is counter intuitive to set a default value, considering
the toolchain required to build this architecture is well known.
It is especially weird when using a native toolchain and requiring to
unset this variable on the command line.

Signed-off-by: Thomas Monjalon 
---
 mk/machine/thunderx/rte.vars.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mk/machine/thunderx/rte.vars.mk b/mk/machine/thunderx/rte.vars.mk
index 81da1b4..ad5a379 100644
--- a/mk/machine/thunderx/rte.vars.mk
+++ b/mk/machine/thunderx/rte.vars.mk
@@ -47,7 +47,7 @@
 #
 
 # ARCH =
-CROSS ?= aarch64-thunderx-linux-gnu-
+# CROSS =
 # MACHINE_CFLAGS =
 # MACHINE_LDFLAGS =
 # MACHINE_ASFLAGS =
-- 
2.7.0

Re: [dpdk-dev] [PATCH] mk: remove default toolchain prefix for ThunderX

[Jerin Jacob]

On Mon, Jan 23, 2017 at 10:15:57AM +0100, Thomas Monjalon wrote:
> The environment variable CROSS must be set when using a cross-toolchain.
> However it is counter intuitive to set a default value, considering
> the toolchain required to build this architecture is well known.
> It is especially weird when using a native toolchain and requiring to
> unset this variable on the command line.
> 
> Signed-off-by: Thomas Monjalon 

Reviewed-by: Jerin Jacob 

> ---
>  mk/machine/thunderx/rte.vars.mk | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mk/machine/thunderx/rte.vars.mk b/mk/machine/thunderx/rte.vars.mk
> index 81da1b4..ad5a379 100644
> --- a/mk/machine/thunderx/rte.vars.mk
> +++ b/mk/machine/thunderx/rte.vars.mk
> @@ -47,7 +47,7 @@
>  #
>  
>  # ARCH =
> -CROSS ?= aarch64-thunderx-linux-gnu-
> +# CROSS =
>  # MACHINE_CFLAGS =
>  # MACHINE_LDFLAGS =
>  # MACHINE_ASFLAGS =
> -- 
> 2.7.0
> 

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Ferruh Yigit]

On 1/22/2017 2:45 AM, Yuanhan Liu wrote:
> On Fri, Jan 20, 2017 at 03:27:43PM +, Ferruh Yigit wrote:
>> On 1/20/2017 11:21 AM, Ferruh Yigit wrote:
>>> On 1/20/2017 8:04 AM, Yuanhan Liu wrote:
 Fix an silly error by auto-complete while managing the merge conflicts.
 It's the eth_dev_data (but not eth_dev) entry should be memset.

 Fixes: d948f596fee2 ("ethdev: fix port data mismatched in multiple process 
 model")

 Reported-by: Ferruh Yigit 
 Signed-off-by: Yuanhan Liu 
 ---
  lib/librte_ether/rte_ethdev.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

 diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
 index 4790faf..61f44e2 100644
 --- a/lib/librte_ether/rte_ethdev.c
 +++ b/lib/librte_ether/rte_ethdev.c
 @@ -225,7 +225,7 @@ struct rte_eth_dev *
return NULL;
}
  
 -  memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
 +  memset(&rte_eth_dev_data[port_id], 0, sizeof(struct rte_eth_dev_data));
>>>
>>> Not directly related to the this issue, but, after fix, this may have
>>> issues with secondary process.
>>>
>>> There were patches sent to fix this.
>>
>> I mean this one:
>> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> 
> d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> model") should have fixed it.

Think about case, where secondary process uses a virtual PMD, which does
a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
device data?

> 
>   --yliu
> 

[dpdk-dev] [PATCH 2/2] net/i40e: fix deletion of all macvlan filters

[Jingjing Wu]

filter_type is not set when removing all macvlan filters. It will
cause error when send AQ command to HW.
This patch fixes this issue.

Fixes: 4861cde46116 ("i40e: new poll mode driver")

CC:sta...@dpdk.org
Signed-off-by: Jingjing Wu 
---
 drivers/net/i40e/i40e_ethdev.c | 5 -
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 92c750f..a818998 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -6115,7 +6115,7 @@ i40e_find_all_mac_for_vlan(struct i40e_vsi *vsi,
 static int
 i40e_vsi_remove_all_macvlan_filter(struct i40e_vsi *vsi)
 {
-   int i, num;
+   int i, j, num;
struct i40e_mac_filter *f;
struct i40e_macvlan_filter *mv_f;
int ret = I40E_SUCCESS;
@@ -6140,6 +6140,7 @@ i40e_vsi_remove_all_macvlan_filter(struct i40e_vsi *vsi)
TAILQ_FOREACH(f, &vsi->mac_list, next) {
(void)rte_memcpy(&mv_f[i].macaddr,
&f->mac_info.mac_addr, ETH_ADDR_LEN);
+   mv_f[i].filter_type = f->mac_info.filter_type;
mv_f[i].vlan_id = 0;
i++;
}
@@ -6149,6 +6150,8 @@ i40e_vsi_remove_all_macvlan_filter(struct i40e_vsi *vsi)
vsi->vlan_num, &f->mac_info.mac_addr);
if (ret != I40E_SUCCESS)
goto DONE;
+   for (j = i; j < i + vsi->vlan_num; j++)
+   mv_f[j].filter_type = f->mac_info.filter_type;
i += vsi->vlan_num;
}
}
-- 
2.4.11

[dpdk-dev] [PATCH 1/2] net/i40e: fix MAC check when delete MAC

[Jingjing Wu]

When VF sends request to remove MAC address, PF host will check
if it is a non-zero or unicast address. When VF remove a multicast
address, it will report error.
This patch fixes this issue.

Fixes: ec852c94af39 ("net/i40e: enhance sanity check of MAC")
Signed-off-by: Jingjing Wu 
---
 drivers/net/i40e/i40e_pf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/i40e/i40e_pf.c b/drivers/net/i40e/i40e_pf.c
index dbea5d0..0af0573 100644
--- a/drivers/net/i40e/i40e_pf.c
+++ b/drivers/net/i40e/i40e_pf.c
@@ -929,7 +929,7 @@ i40e_pf_host_process_cmd_del_ether_address(struct 
i40e_pf_vf *vf,
 
for (i = 0; i < addr_list->num_elements; i++) {
mac = (struct ether_addr *)(addr_list->list[i].addr);
-   if(!is_valid_assigned_ether_addr(mac) ||
+   if(is_zero_ether_addr(mac) ||
i40e_vsi_delete_mac(vf->vsi, mac)) {
ret = I40E_ERR_INVALID_MAC_ADDR;
goto send_msg;
-- 
2.4.11

Re: [dpdk-dev] [PATCH v2] net/sfc: do not panic if alarms are not supported

[Ferruh Yigit]

On 1/19/2017 11:12 AM, Andrew Rybchenko wrote:
> Alarms are not supported on the FreeBSD.
> Application must poll link status periodically itself using
> rte_eth_link_get_nowait() to avoid management event queue overflow.
> 
> Fixes: 2de39f4e1310 ("net/sfc: periodic management EVQ polling using alarm")
> 
> Signed-off-by: Andrew Rybchenko 
> Reviewed-by: Andrew Lee 
> Reviewed-by: Andy Moreton 

Applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH v3] net/sfc: enable TSO by default

[Ferruh Yigit]

On 1/20/2017 3:22 PM, Andrew Rybchenko wrote:
> Remove RTE_LIBRTE_SFC_EFX_TSO config option since it is not
> required any more:
>  - unreasonable limit on number of Tx queues when TSO is not
>actually required should be solved using per-device parameter
>  - performance difference with and without TSO compiled in is small
> 
> Signed-off-by: Andrew Rybchenko 

Applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH] net/sfc: fix TSO limits imposed to the number of Tx queues

[Ferruh Yigit]

On 1/19/2017 10:46 AM, Andrew Rybchenko wrote:
> From: Ivan Malov 
> 
> The number of Tx queues requested by the user must not be overridden;
> instead, the limits imposed by TSO must be applied to the advertised
> maximum
> 
> Fixes: fec33d5bb3eb ("net/sfc: support firmware-assisted TSO")
> 
> Signed-off-by: Ivan Malov 
> Signed-off-by: Andrew Rybchenko 
> Reviewed-by: Andrew Lee 
> Reviewed-by: Andy Moreton 

Can you please rebase this patch on top of latest next-net, since
another tso patch [1] applied first.

[1]
http://dpdk.org/dev/patchwork/patch/19842/

Thanks,
ferruh

Re: [dpdk-dev] [PATCH 1/2] net/sfc: use eal I/O device memory read/write API

[Ferruh Yigit]

On 1/20/2017 1:53 PM, Andrew Rybchenko wrote:
> Use relaxed version of these functions to guarantee no changes on
> the step.
> 
> Signed-off-by: Andrew Rybchenko 

Series applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH] net/qede/base: fix freebsd compilation

[Thomas Monjalon]

2017-01-22 21:02, Rasesh Mody:
> This patch addresses compilation errors on FreeBSD 11.0 clang 3.8.0.

Is it really specific to FreeBSD 11? I guess it is more about clang.

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Yuanhan Liu]

On Mon, Jan 23, 2017 at 09:41:35AM +, Ferruh Yigit wrote:
> On 1/22/2017 2:45 AM, Yuanhan Liu wrote:
> > On Fri, Jan 20, 2017 at 03:27:43PM +, Ferruh Yigit wrote:
> >> On 1/20/2017 11:21 AM, Ferruh Yigit wrote:
> >>> On 1/20/2017 8:04 AM, Yuanhan Liu wrote:
>  Fix an silly error by auto-complete while managing the merge conflicts.
>  It's the eth_dev_data (but not eth_dev) entry should be memset.
> 
>  Fixes: d948f596fee2 ("ethdev: fix port data mismatched in multiple 
>  process model")
> 
>  Reported-by: Ferruh Yigit 
>  Signed-off-by: Yuanhan Liu 
>  ---
>   lib/librte_ether/rte_ethdev.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
>  diff --git a/lib/librte_ether/rte_ethdev.c 
>  b/lib/librte_ether/rte_ethdev.c
>  index 4790faf..61f44e2 100644
>  --- a/lib/librte_ether/rte_ethdev.c
>  +++ b/lib/librte_ether/rte_ethdev.c
>  @@ -225,7 +225,7 @@ struct rte_eth_dev *
>   return NULL;
>   }
>   
>  -memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
>  +memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
>  rte_eth_dev_data));
> >>>
> >>> Not directly related to the this issue, but, after fix, this may have
> >>> issues with secondary process.
> >>>
> >>> There were patches sent to fix this.
> >>
> >> I mean this one:
> >> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > 
> > d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> > model") should have fixed it.
> 
> Think about case, where secondary process uses a virtual PMD, which does
> a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
> device data?

Yes, it may. However, I doubt that's the typical usage. Besides that,
most of virtual PMDs don't support Multipleprocess: git grep shows pcap
is the only one that does claim Multipleprocess is supported.

--yliu

Re: [dpdk-dev] [PATCH 0/3] few virtio/vhost fixes

[Yuanhan Liu]

On Sun, Jan 22, 2017 at 04:46:57PM +0800, Yuanhan Liu wrote:
> ---
> Yuanhan Liu (3):
>   vhost: fix dead loop in enqueue path
>   vhost: fix long stall of vhost-user negotiation
>   net/virtio: fix crash when number of virtio devices > 1

Applied to dpdk-next-virtio.

--yliu

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Ferruh Yigit]

On 1/23/2017 10:34 AM, Yuanhan Liu wrote:
> On Mon, Jan 23, 2017 at 09:41:35AM +, Ferruh Yigit wrote:
>> On 1/22/2017 2:45 AM, Yuanhan Liu wrote:
>>> On Fri, Jan 20, 2017 at 03:27:43PM +, Ferruh Yigit wrote:
 On 1/20/2017 11:21 AM, Ferruh Yigit wrote:
> On 1/20/2017 8:04 AM, Yuanhan Liu wrote:
>> Fix an silly error by auto-complete while managing the merge conflicts.
>> It's the eth_dev_data (but not eth_dev) entry should be memset.
>>
>> Fixes: d948f596fee2 ("ethdev: fix port data mismatched in multiple 
>> process model")
>>
>> Reported-by: Ferruh Yigit 
>> Signed-off-by: Yuanhan Liu 
>> ---
>>  lib/librte_ether/rte_ethdev.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/lib/librte_ether/rte_ethdev.c 
>> b/lib/librte_ether/rte_ethdev.c
>> index 4790faf..61f44e2 100644
>> --- a/lib/librte_ether/rte_ethdev.c
>> +++ b/lib/librte_ether/rte_ethdev.c
>> @@ -225,7 +225,7 @@ struct rte_eth_dev *
>>  return NULL;
>>  }
>>  
>> -memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
>> +memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
>> rte_eth_dev_data));
>
> Not directly related to the this issue, but, after fix, this may have
> issues with secondary process.
>
> There were patches sent to fix this.

 I mean this one:
 http://dpdk.org/ml/archives/dev/2017-January/054422.html
>>>
>>> d948f596fee2 ("ethdev: fix port data mismatched in multiple process
>>> model") should have fixed it.
>>
>> Think about case, where secondary process uses a virtual PMD, which does
>> a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
>> device data?
> 
> Yes, it may. However, I doubt that's the typical usage. 

But this is a use case, and broken now, and fix is known. Should be
fixed I think.

> Besides that,
> most of virtual PMDs don't support Multipleprocess: git grep shows pcap
> is the only one that does claim Multipleprocess is supported.

I guess you searched for NIC feature documentation for this. But as far
as I know, all virtual drivers can be used in both primary and secondary
process.

> 
>   --yliu
> 

[dpdk-dev] [PATCH v2] net/sfc: fix TSO limits imposed to the number of Tx queues

[Andrew Rybchenko]

From: Ivan Malov 

The number of Tx queues requested by the user must not be overridden;
instead, the limits imposed by TSO must be applied to the advertised
maximum

Fixes: fec33d5bb3eb ("net/sfc: support firmware-assisted TSO")

Signed-off-by: Ivan Malov 
Signed-off-by: Andrew Rybchenko 
Reviewed-by: Andrew Lee 
Reviewed-by: Andy Moreton 
---
v2:
 * rebase becase of net/sfc: enable TSO by default

 drivers/net/sfc/sfc.c| 18 +-
 drivers/net/sfc/sfc_tx.c |  5 -
 2 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/drivers/net/sfc/sfc.c b/drivers/net/sfc/sfc.c
index 457a53e..03ecec2 100644
--- a/drivers/net/sfc/sfc.c
+++ b/drivers/net/sfc/sfc.c
@@ -203,6 +203,13 @@
limits.edl_max_txq_count =
MIN(encp->enc_txq_limit,
limits.edl_max_evq_count - 1 - limits.edl_max_rxq_count);
+
+   if (sa->tso)
+   limits.edl_max_txq_count =
+   MIN(limits.edl_max_txq_count,
+   encp->enc_fw_assisted_tso_v2_n_contexts /
+   encp->enc_hw_pf_count);
+
SFC_ASSERT(limits.edl_max_txq_count >= limits.edl_min_rxq_count);
 
/* Configure the minimum required resources needed for the
@@ -601,12 +608,17 @@
if (rc != 0)
goto fail_nic_reset;
 
+   encp = efx_nic_cfg_get(sa->nic);
+
+   sa->tso = encp->enc_fw_assisted_tso_v2_enabled;
+   if (!sa->tso)
+   sfc_warn(sa, "TSO support isn't available on this adapter");
+
sfc_log_init(sa, "estimate resource limits");
rc = sfc_estimate_resource_limits(sa);
if (rc != 0)
goto fail_estimate_rsrc_limits;
 
-   encp = efx_nic_cfg_get(sa->nic);
sa->txq_max_entries = encp->enc_txq_max_ndescs;
SFC_ASSERT(rte_is_power_of_2(sa->txq_max_entries));
 
@@ -621,10 +633,6 @@
if (rc != 0)
goto fail_set_rss_defaults;
 
-   sa->tso = efx_nic_cfg_get(sa->nic)->enc_fw_assisted_tso_v2_enabled;
-   if (!sa->tso)
-   sfc_warn(sa, "TSO support isn't available on this adapter");
-
sfc_log_init(sa, "fini nic");
efx_nic_fini(enp);
 
diff --git a/drivers/net/sfc/sfc_tx.c b/drivers/net/sfc/sfc_tx.c
index e772584..5a6282c 100644
--- a/drivers/net/sfc/sfc_tx.c
+++ b/drivers/net/sfc/sfc_tx.c
@@ -312,11 +312,6 @@
 
sa->txq_count = sa->eth_dev->data->nb_tx_queues;
 
-   if (sa->tso)
-   sa->txq_count = MIN(sa->txq_count,
-  efx_nic_cfg_get(sa->nic)->enc_fw_assisted_tso_v2_n_contexts /
-  efx_nic_cfg_get(sa->nic)->enc_hw_pf_count);
-
sa->txq_info = rte_calloc_socket("sfc-txqs", sa->txq_count,
 sizeof(sa->txq_info[0]), 0,
 sa->socket_id);
-- 
1.8.2.3

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Yuanhan Liu]

On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
> >>  lib/librte_ether/rte_ethdev.c | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/lib/librte_ether/rte_ethdev.c 
> >> b/lib/librte_ether/rte_ethdev.c
> >> index 4790faf..61f44e2 100644
> >> --- a/lib/librte_ether/rte_ethdev.c
> >> +++ b/lib/librte_ether/rte_ethdev.c
> >> @@ -225,7 +225,7 @@ struct rte_eth_dev *
> >>return NULL;
> >>}
> >>  
> >> -  memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
> >> +  memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
> >> rte_eth_dev_data));
> >
> > Not directly related to the this issue, but, after fix, this may have
> > issues with secondary process.
> >
> > There were patches sent to fix this.
> 
>  I mean this one:
>  http://dpdk.org/ml/archives/dev/2017-January/054422.html
> >>>
> >>> d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> >>> model") should have fixed it.
> >>
> >> Think about case, where secondary process uses a virtual PMD, which does
> >> a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
> >> device data?
> > 
> > Yes, it may. However, I doubt that's the typical usage. 
> 
> But this is a use case, and broken now,

I thought it was broken since the beginning?

> and fix is known.

And there is already a fix?

> Should be
> fixed I think.

Sure.

> 
> > Besides that,
> > most of virtual PMDs don't support Multipleprocess: git grep shows pcap
> > is the only one that does claim Multipleprocess is supported.
> 
> I guess you searched for NIC feature documentation for this.

Yes.

> But as far
> as I know, all virtual drivers can be used in both primary and secondary
> process.

Maybe. But it becomes very error-prone to me then when vdev are involved
in both primary and secondary process. I don't think current code is (or
designed to be) strong enough to support that.

I don't know it's allowed to use hotplug or not in the multiple process
model. If yes, I think there would be many ways to break it.

Honestly, the multiple process doesn't look like a good/clean design to
me, especially when some piece of code claim to support it while some
other doesn't.

So my point was, yes, there is a bug, we should fix it. But it seems
that there could be so many bugs if we hugely expand the test coverage
of the multiple process feature.

--yliu

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Ferruh Yigit]

On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
> On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
  lib/librte_ether/rte_ethdev.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

 diff --git a/lib/librte_ether/rte_ethdev.c 
 b/lib/librte_ether/rte_ethdev.c
 index 4790faf..61f44e2 100644
 --- a/lib/librte_ether/rte_ethdev.c
 +++ b/lib/librte_ether/rte_ethdev.c
 @@ -225,7 +225,7 @@ struct rte_eth_dev *
return NULL;
}
  
 -  memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
 +  memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
 rte_eth_dev_data));
>>>
>>> Not directly related to the this issue, but, after fix, this may have
>>> issues with secondary process.
>>>
>>> There were patches sent to fix this.
>>
>> I mean this one:
>> http://dpdk.org/ml/archives/dev/2017-January/054422.html
>
> d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> model") should have fixed it.

 Think about case, where secondary process uses a virtual PMD, which does
 a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
 device data?
>>>
>>> Yes, it may. However, I doubt that's the typical usage. 
>>
>> But this is a use case, and broken now,
> 
> I thought it was broken since the beginning?

No, memset(&rte_eth_dev_data[port_id], ...) breaks it.

> 
>> and fix is known.
> 
> And there is already a fix?

http://dpdk.org/ml/archives/dev/2017-January/054422.html

> 
>> Should be
>> fixed I think.
> 
> Sure.
> 
>>
>>> Besides that,
>>> most of virtual PMDs don't support Multipleprocess: git grep shows pcap
>>> is the only one that does claim Multipleprocess is supported.
>>
>> I guess you searched for NIC feature documentation for this.
> 
> Yes.
> 
>> But as far
>> as I know, all virtual drivers can be used in both primary and secondary
>> process.
> 
> Maybe. But it becomes very error-prone to me then when vdev are involved
> in both primary and secondary process. I don't think current code is (or
> designed to be) strong enough to support that.
> 
> I don't know it's allowed to use hotplug or not in the multiple process
> model. If yes, I think there would be many ways to break it.
> 
> Honestly, the multiple process doesn't look like a good/clean design to
> me, especially when some piece of code claim to support it while some
> other doesn't.
> 
> So my point was, yes, there is a bug, we should fix it. But it seems
> that there could be so many bugs if we hugely expand the test coverage
> of the multiple process feature.

Agreed.

> 
>   --yliu
> 

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Yuanhan Liu]

On Mon, Jan 23, 2017 at 11:32:23AM +, Ferruh Yigit wrote:
> On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
> > On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
>   lib/librte_ether/rte_ethdev.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
>  diff --git a/lib/librte_ether/rte_ethdev.c 
>  b/lib/librte_ether/rte_ethdev.c
>  index 4790faf..61f44e2 100644
>  --- a/lib/librte_ether/rte_ethdev.c
>  +++ b/lib/librte_ether/rte_ethdev.c
>  @@ -225,7 +225,7 @@ struct rte_eth_dev *
>   return NULL;
>   }
>   
>  -memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
>  +memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
>  rte_eth_dev_data));
> >>>
> >>> Not directly related to the this issue, but, after fix, this may have
> >>> issues with secondary process.
> >>>
> >>> There were patches sent to fix this.
> >>
> >> I mean this one:
> >> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> >
> > d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> > model") should have fixed it.
> 
>  Think about case, where secondary process uses a virtual PMD, which does
>  a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
>  device data?
> >>>
> >>> Yes, it may. However, I doubt that's the typical usage. 
> >>
> >> But this is a use case, and broken now,
> > 
> > I thought it was broken since the beginning?
> 
> No, memset(&rte_eth_dev_data[port_id], ...) breaks it.

Oh, you were talking about that particular case Remy's patch meant to
fix.

> >> and fix is known.
> > 
> > And there is already a fix?
> 
> http://dpdk.org/ml/archives/dev/2017-January/054422.html

Yes, it should fix that issue. One question: do Remy or you regularly
run some multiple process test cases (and with vdev both in primary
and secondary process)?

--yliu

Re: [dpdk-dev] [PATCH v2] net/sfc: fix TSO limits imposed to the number of Tx queues

[Ferruh Yigit]

On 1/23/2017 11:06 AM, Andrew Rybchenko wrote:
> From: Ivan Malov 
> 
> The number of Tx queues requested by the user must not be overridden;
> instead, the limits imposed by TSO must be applied to the advertised
> maximum
> 
> Fixes: fec33d5bb3eb ("net/sfc: support firmware-assisted TSO")
> 
> Signed-off-by: Ivan Malov 
> Signed-off-by: Andrew Rybchenko 
> Reviewed-by: Andrew Lee 
> Reviewed-by: Andy Moreton 

Applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH v4] doc: add known uio_pci_generic issue for i40e

[Ferruh Yigit]

On 1/23/2017 8:04 AM, Mcnamara, John wrote:
> 
> 
>> -Original Message-
>> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Jeff Guo
>> Sent: Sunday, January 22, 2017 2:04 AM
>> To: Zhang, Helin ; Wu, Jingjing
>> 
>> Cc: dev@dpdk.org; Guo, Jia 
>> Subject: [dpdk-dev] [PATCH v4] doc: add known uio_pci_generic issue for
>> i40e
>>
>> When bind the ``uio_pci_generic`` module in X710/XL710/XXV710, the result
>> is failed. The ``uio_pci_generic`` module is not supported by
>> X710/XL710/XXV710.
>>
>> Signed-off-by: Jeff Guo 
> 
> Acked-by: John McNamara 

Applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Yuanhan Liu]

On Mon, Jan 23, 2017 at 07:40:50PM +0800, Yuanhan Liu wrote:
> On Mon, Jan 23, 2017 at 11:32:23AM +, Ferruh Yigit wrote:
> > On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
> > > On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
> >   lib/librte_ether/rte_ethdev.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> >  diff --git a/lib/librte_ether/rte_ethdev.c 
> >  b/lib/librte_ether/rte_ethdev.c
> >  index 4790faf..61f44e2 100644
> >  --- a/lib/librte_ether/rte_ethdev.c
> >  +++ b/lib/librte_ether/rte_ethdev.c
> >  @@ -225,7 +225,7 @@ struct rte_eth_dev *
> > return NULL;
> > }
> >   
> >  -  memset(&rte_eth_devices[port_id], 0, sizeof(*eth_dev->data));
> >  +  memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
> >  rte_eth_dev_data));
> > >>>
> > >>> Not directly related to the this issue, but, after fix, this may 
> > >>> have
> > >>> issues with secondary process.
> > >>>
> > >>> There were patches sent to fix this.
> > >>
> > >> I mean this one:
> > >> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > >
> > > d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> > > model") should have fixed it.
> > 
> >  Think about case, where secondary process uses a virtual PMD, which 
> >  does
> >  a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
> >  device data?
> > >>>
> > >>> Yes, it may. However, I doubt that's the typical usage. 
> > >>
> > >> But this is a use case, and broken now,
> > > 
> > > I thought it was broken since the beginning?
> > 
> > No, memset(&rte_eth_dev_data[port_id], ...) breaks it.
> 
> Oh, you were talking about that particular case Remy's patch meant to
> fix.
> 
> > >> and fix is known.
> > > 
> > > And there is already a fix?
> > 
> > http://dpdk.org/ml/archives/dev/2017-January/054422.html
> 
> Yes, it should fix that issue.

Well, few more thoughts: it may fix the crash issue Remy saw, but it
looks like more a workaround to me. Basically, if primary and secondary
shares a same port id, they should point to same device. Otherwise,
primary process may use eth_dev->data for a device A, while the
secondary process may use it for another device, as you said, it
could be a vdev.

In such case, there is no way we could continue safely. That said,
the given patch avoids the total reset of eth_dev->data, while it
continues reset the eth_dev->data->name, which is wrong.

So it's not a proper fix.

Again, I think it's more about the usage. If primary starts with
a nic device A, while the secondary starts with a nic device B,
there is no way they could work well (unless they use different
port id).

--yliu

> One question: do Remy or you regularly
> run some multiple process test cases (and with vdev both in primary
> and secondary process)?

[dpdk-dev] [PATCHv6 00/33] NXP DPAA2 PMD

[Hemant Agrawal]

The patch series adds NXP’s QorIQ-Layerscape DPAA2 Architecture based
fsl-mc bus driver and network SoC PMD.  This version of the driver
supports NXP LS208xA, LS204xA and LS108x families Network SoCs.

DPAA2, or Data Path Acceleration Architecture, is a hardware architecture
designed for high-speed network packet processing. It uses a bus name
‘fsl-mc’, part of Linux Kernel Staging tree [3], for resource management.

A brief description of architecture is given below; detailed description
is part of the documentation in the patches itself.

DPAA2 contains hardware component called the Management Complex (or MC).
It manages the DPAA2 hardware resources.  The MC provides an object-based
abstraction for software drivers to use the DPAA2 hardware.

Some of the key objects are:
- DPNI, which refers to the network interface object.
- DPBP, which refers to HW based memory pool object
- DPIO, refers to processing context for accessing QBMAN

Besides the MC, DPAA2 also includes a Hardware based Queue and Buffer Manager
called QBMAN. Prime responsibility of QBMAN is to allow lockless access to
software/user-space to the queues and buffers implemented in the hardware.

The patch series could be logically structured into following sub-areas:
1. (Patch 0001) Enabling crc in armv8 core machine type
2. (Patch 0002) Common dpaa2 hw accelerator drivers for QBMAN.
3. (Patches 0003-0017) introduce fsl-mc bus, dpaa2 pmd, DPIO and mempool
4. (Patches 0018-0031) Support for DPAA2 Ethernet Device (ethdev)
5. (Patches 0032-0033) Additional functionality in DPAA2 ethdev.

The following design decisions are made during development:

1. DPAA2 implements a new bus called "fsl-mc" and some common accelerator 
drivers.
   These drivers will be shared with dpaa2 based crypto drivers.

2. DPAA2 implements the HW mempool offload with DPBP object.
 - The new pool is being configured using compile time option and pool name
   as "dpaa2".

3. It maintains per lcore DPIO objects and affine the DPIO instance to the
   processing threads accessing the QBMAN HW.

Prerequisites:
 - For running the PMD, NXP's SoC (board) and SDK (software/BSP) is required.
   Information about obtaining relevant software is available in the docs
   as part of the patch.
 - At present the series has limited support for Ethernet functions. But,
   more functionality would be made available in a phased manner.

Future Changes/Caveats:

1. VFIO code for fsl-mc bus is different than eal-vfio code for pci bus.
   This need to be re-worked to make possible re-use of the existing code.

2. shared lib support in case of parallel build fails due to dependency of
   drivers/common, pool, etc on other driver components. The dependency graph
   need to be improved to support it.

References:
[1] http://dpdk.org/dev/patchwork/patch/19557/
[2] https://www.kernel.org/doc/readme/drivers-staging-fsl-mc-README.txt
[3] http://dpdk.org/ml/archives/dev/2016-October/048949.html

---
v6:
* rebased over master (61207d0)
* removing DPAA2_COMMON as configurable option
* renaming drivers bus, pool libraries removing 'pmd'
* Headers of Licenses
* exposed variable renaming with *rte_*  prefix
* handling Ferruh's comment for NXP dpaa2 driver
* moving around MAINTAINER and DOC file patches 

v5:
* rebased over master (6818a7f4)

v4:
* rebased over master (1feda4d8) and patches from Shreyansh [1] for Bus Arch.

v3:
* rebased over master (eac901ce2) and patches from Shreyansh [1] for Bus Arch.
* Fixed comment from John on Patch-0003 for documentation
* Removed Patch-0001 for rte_device in rte_eth_dev; Already upstreamed through
  another series

v2:
* separated the "fsl-mc" bus from the dpaa2 pmd driver - introduced drivers/bus
* separated the "dpaa2" hw mempool from dpaa2 pmd driver - introduced 
drivers/pool
* removed documentation warnings and missing information.
* removed arm64 part specific code from driver
* changed rte_panic to errors
* reduced checkpatch warnings


Hemant Agrawal (33):
  mk/dpaa2: add the crc support to the machine type
  drivers/common/dpaa2: adding qbman driver
  bus/fslmc: introducing fsl-mc bus driver
  bus/fslmc: introduce mc object functions
  bus/fslmc: add mc dpni object support
  bus/fslmc: add mc dpio object support
  bus/fslmc: add mc dpbp object support
  bus/fslmc: add mc dpseci object support
  eal/vfio: adding vfio utility functions in map file
  bus/fslmc: add vfio support
  bus/fslmc: scan for net and sec devices
  net/dpaa2: introducing NXP dpaa2 pmd driver
  doc: add dpaa2 nic details
  bus/fslmc: add debug log message support
  drivers/common/dpaa2: dpio portal driver
  drivers/pool/dpaa2: adding hw offloaded mempool
  drivers/common/dpaa2: dpio routine to affine to crypto threads
  net/dpaa2: adding eth ops to dpaa2
  net/dpaa2: add rss flow distribution
  net/dpaa2: configure mac address at init
  net/dpaa2: attach the buffer pool to dpni
  net/dpaa2: add support for l3 and l4 checksum offload
  net/dpaa2: add support for promiscuous mode

[dpdk-dev] [PATCHv6 01/33] mk/dpaa2: add the crc support to the machine type

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
Acked-by: Jerin Jacob 
---
 mk/machine/dpaa2/rte.vars.mk | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/mk/machine/dpaa2/rte.vars.mk b/mk/machine/dpaa2/rte.vars.mk
index 8541633..e4735c2 100644
--- a/mk/machine/dpaa2/rte.vars.mk
+++ b/mk/machine/dpaa2/rte.vars.mk
@@ -1,6 +1,7 @@
 #   BSD LICENSE
 #
-#   Copyright(c) 2016 Freescale Semiconductor, Inc. All rights reserved.
+#   Copyright (c) 2016 Freescale Semiconductor, Inc. All rights reserved.
+#   Copyright (c) 2016 NXP. All rights reserved.
 #
 #   Redistribution and use in source and binary forms, with or without
 #   modification, are permitted provided that the following conditions
@@ -53,7 +54,7 @@
 # CPU_CFLAGS =
 # CPU_LDFLAGS =
 # CPU_ASFLAGS =
-MACHINE_CFLAGS += -march=armv8-a
+MACHINE_CFLAGS += -march=armv8-a+crc
 
 ifdef CONFIG_RTE_ARCH_ARM_TUNE
 MACHINE_CFLAGS += -mcpu=$(CONFIG_RTE_ARCH_ARM_TUNE)
-- 
1.9.1

[dpdk-dev] [PATCHv6 04/33] bus/fslmc: introduce mc object functions

[Hemant Agrawal]

This patch intoduces the DPAA2 MC(Management complex Driver).

This is a minimal set of low level functions to send and
receive commands to the fsl-mc. It includes support for basic
management commands and commands to manipulate MC objects.

This is common to be used by various DPAA2 PMDs. e.g.net, crypto
and other drivers.

This is a low level library also used in kernel.

Signed-off-by: Cristian Sovaiala 
Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile|   7 ++
 drivers/bus/fslmc/mc/fsl_mc_cmd.h | 238 ++
 drivers/bus/fslmc/mc/fsl_mc_sys.h | 105 +
 drivers/bus/fslmc/mc/mc_sys.c | 114 ++
 4 files changed, 464 insertions(+)
 create mode 100644 drivers/bus/fslmc/mc/fsl_mc_cmd.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_mc_sys.h
 create mode 100644 drivers/bus/fslmc/mc/mc_sys.c

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 5a0fd61..1058f0f 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -37,6 +37,10 @@ LIB = librte_bus_fslmc.a
 
 CFLAGS += -O3
 CFLAGS += $(WERROR_FLAGS)
+CFLAGS += "-Wno-strict-aliasing"
+
+CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
+CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/mc
 
 # versioning export map
 EXPORT_MAP := rte_bus_fslmc_version.map
@@ -44,6 +48,9 @@ EXPORT_MAP := rte_bus_fslmc_version.map
 # library version
 LIBABIVER := 1
 
+SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
+mc/mc_sys.c
+
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
 
 # library dependencies
diff --git a/drivers/bus/fslmc/mc/fsl_mc_cmd.h 
b/drivers/bus/fslmc/mc/fsl_mc_cmd.h
new file mode 100644
index 000..bc41646
--- /dev/null
+++ b/drivers/bus/fslmc/mc/fsl_mc_cmd.h
@@ -0,0 +1,238 @@
+/*-
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ *   BSD LICENSE
+ *
+ * Copyright 2013-2016 Freescale Semiconductor Inc.
+ * Copyright (c) 2016 NXP.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *   GPL LICENSE SUMMARY
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef __FSL_MC_CMD_H
+#define __FSL_MC_CMD_H
+
+#define MC_CMD_NUM_OF_PARAMS   7
+
+#define MAKE_UMASK64(_width) \
+   ((uint64_t)((_width) < 64 ? ((uint64_t)1 << (_width)) - 1 : \
+  (uint64_t)-1))
+
+static inline uint64_t mc_enc(int lsoffset, int width, uint64_t val)
+{
+   return (uint64_t)(((uint64_t)val & MAKE_UMASK64(width)) << lsoffset);
+}
+
+static inline uint64_t mc_dec(uint64_t val, int lsoffset, int width)
+{
+   return (uint64_t)((val >> lsoffset) & MAKE_UMASK64(width));
+}
+
+struct mc_command {
+   uint64_t header;
+   uint64_t params[MC_CMD_NUM_OF_PARAMS];
+};
+
+/**
+ * enum mc_cmd_status - indicates MC status at command response
+ * @MC_CMD_STATUS_OK: Completed successfully
+ * @MC_CMD_STATUS_READY: Ready to be processed
+ * @MC_CMD_STATUS_AUTH_ERR: Authentication error
+ * @MC_CMD_STATUS_NO_PRIVILEGE: No privilege
+ * @MC_CMD_STATUS_DMA_ERR: DMA or I/O error
+ * @MC_CMD_STATUS_CONFIG_ERR: Configuration error
+ * @MC_CMD_STATUS_TIMEOUT: Operation timed out
+ * @MC_CMD_STATUS_NO_RESOURCE: No resources
+ * @MC_CMD_STATUS_NO_MEMORY: No memory available
+ * @MC_CMD_STATUS_BUSY: Device is busy
+ * @MC_

[dpdk-dev] [PATCHv6 03/33] bus/fslmc: introducing fsl-mc bus driver

[Hemant Agrawal]

The fslmc bus driver is a rte_bus driver which scans the fsl-mc bus
for NXP DPAA2 SoCs.

Signed-off-by: Hemant Agrawal 
---
 MAINTAINERS |   1 +
 config/common_base  |   5 +
 config/defconfig_arm64-dpaa2-linuxapp-gcc   |   5 +
 drivers/Makefile|   1 +
 drivers/bus/Makefile|  36 +++
 drivers/bus/fslmc/Makefile  |  52 ++
 drivers/bus/fslmc/fslmc_bus.c   | 125 +++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   7 ++
 drivers/bus/fslmc/rte_fslmc.h   | 148 
 drivers/common/Makefile |   4 +
 drivers/common/dpaa2/Makefile   |   4 +
 drivers/common/dpaa2/qbman/Makefile |   4 +
 12 files changed, 392 insertions(+)
 create mode 100644 drivers/bus/Makefile
 create mode 100644 drivers/bus/fslmc/Makefile
 create mode 100644 drivers/bus/fslmc/fslmc_bus.c
 create mode 100644 drivers/bus/fslmc/rte_bus_fslmc_version.map
 create mode 100644 drivers/bus/fslmc/rte_fslmc.h

diff --git a/MAINTAINERS b/MAINTAINERS
index 5ad150d..4a13140 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -356,6 +356,7 @@ F: doc/guides/nics/nfp.rst
 
 NXP dpaa2
 M: Hemant Agrawal 
+F: drivers/bus/fslmc/
 F: drivers/common/dpaa2/
 
 QLogic bnx2x
diff --git a/config/common_base b/config/common_base
index 5564d92..bf1de8f 100644
--- a/config/common_base
+++ b/config/common_base
@@ -287,6 +287,11 @@ CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_TX=n
 CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_DRIVER=n
 CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_MBOX=n
 
+#
+# Compile NXP DPAA2 FSL-MC Bus
+#
+CONFIG_RTE_LIBRTE_FSLMC_BUS=n
+
 # Compile burst-oriented VIRTIO PMD driver
 #
 CONFIG_RTE_LIBRTE_VIRTIO_PMD=y
diff --git a/config/defconfig_arm64-dpaa2-linuxapp-gcc 
b/config/defconfig_arm64-dpaa2-linuxapp-gcc
index da23bab..365ae5a 100644
--- a/config/defconfig_arm64-dpaa2-linuxapp-gcc
+++ b/config/defconfig_arm64-dpaa2-linuxapp-gcc
@@ -41,3 +41,8 @@ CONFIG_RTE_ARCH_ARM_TUNE="cortex-a57+fp+simd"
 #
 CONFIG_RTE_MAX_LCORE=8
 CONFIG_RTE_MAX_NUMA_NODES=1
+
+#
+# Compile NXP DPAA2 FSL-MC Bus
+#
+CONFIG_RTE_LIBRTE_FSLMC_BUS=y
diff --git a/drivers/Makefile b/drivers/Makefile
index d5580f6..bdae63b 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -32,6 +32,7 @@
 include $(RTE_SDK)/mk/rte.vars.mk
 
 DIRS-y += common
+DIRS-y += bus
 DIRS-y += net
 DIRS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += crypto
 
diff --git a/drivers/bus/Makefile b/drivers/bus/Makefile
new file mode 100644
index 000..60e9764
--- /dev/null
+++ b/drivers/bus/Makefile
@@ -0,0 +1,36 @@
+#   BSD LICENSE
+#
+#   Copyright(c) 2016 NXP. All rights reserved.
+#   All rights reserved.
+#
+#   Redistribution and use in source and binary forms, with or without
+#   modification, are permitted provided that the following conditions
+#   are met:
+#
+# * Redistributions of source code must retain the above copyright
+#   notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above copyright
+#   notice, this list of conditions and the following disclaimer in
+#   the documentation and/or other materials provided with the
+#   distribution.
+# * Neither the name of NXP nor the names of its
+#   contributors may be used to endorse or promote products derived
+#   from this software without specific prior written permission.
+#
+#   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+#   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+#   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+#   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+#   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+#   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+#   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+#   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+#   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+#   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+DIRS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc
+
+include $(RTE_SDK)/mk/rte.subdir.mk
diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
new file mode 100644
index 000..5a0fd61
--- /dev/null
+++ b/drivers/bus/fslmc/Makefile
@@ -0,0 +1,52 @@
+#   BSD LICENSE
+#
+#   Copyright (c) 2016 NXP. All rights reserved.
+#
+#   Redistribution and use in source and binary forms, with or without
+#   modification, are permitted provided that the following conditions
+#   are met:
+#
+# * Redistributions of source code must retain the above copyright
+#   notice, this list of conditions and the following disclaimer.
+#   

[dpdk-dev] [PATCHv6 05/33] bus/fslmc: add mc dpni object support

[Hemant Agrawal]

This patch add support for dpni object support in MC
driver.

DPNI represent a network interface object in DPAA2.

Signed-off-by: Alex Marginean 
Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile  |1 +
 drivers/bus/fslmc/mc/dpni.c |  739 
 drivers/bus/fslmc/mc/fsl_dpkg.h |  184 
 drivers/bus/fslmc/mc/fsl_dpni.h | 1217 +++
 drivers/bus/fslmc/mc/fsl_dpni_cmd.h |  334 
 drivers/bus/fslmc/mc/fsl_net.h  |  487 +++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   22 +
 7 files changed, 2984 insertions(+)
 create mode 100644 drivers/bus/fslmc/mc/dpni.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpkg.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpni.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpni_cmd.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_net.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 1058f0f..15ab89a 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -49,6 +49,7 @@ EXPORT_MAP := rte_bus_fslmc_version.map
 LIBABIVER := 1
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
+mc/dpni.c \
 mc/mc_sys.c
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
diff --git a/drivers/bus/fslmc/mc/dpni.c b/drivers/bus/fslmc/mc/dpni.c
new file mode 100644
index 000..3330614
--- /dev/null
+++ b/drivers/bus/fslmc/mc/dpni.c
@@ -0,0 +1,739 @@
+/*-
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ *   BSD LICENSE
+ *
+ * Copyright 2013-2016 Freescale Semiconductor Inc.
+ * Copyright (c) 2016 NXP.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *   GPL LICENSE SUMMARY
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#include 
+#include 
+#include 
+#include 
+
+int dpni_prepare_key_cfg(const struct dpkg_profile_cfg *cfg,
+uint8_t *key_cfg_buf)
+{
+   int i, j;
+   int offset = 0;
+   int param = 1;
+   uint64_t *params = (uint64_t *)key_cfg_buf;
+
+   if (!key_cfg_buf || !cfg)
+   return -EINVAL;
+
+   params[0] |= mc_enc(0, 8, cfg->num_extracts);
+   params[0] = cpu_to_le64(params[0]);
+
+   if (cfg->num_extracts >= DPKG_MAX_NUM_OF_EXTRACTS)
+   return -EINVAL;
+
+   for (i = 0; i < cfg->num_extracts; i++) {
+   switch (cfg->extracts[i].type) {
+   case DPKG_EXTRACT_FROM_HDR:
+   params[param] |= mc_enc(0, 8,
+   cfg->extracts[i].extract.from_hdr.prot);
+   params[param] |= mc_enc(8, 4,
+   cfg->extracts[i].extract.from_hdr.type);
+   params[param] |= mc_enc(16, 8,
+   cfg->extracts[i].extract.from_hdr.size);
+   params[param] |= mc_enc(24, 8,
+   cfg->extracts[i].extract.
+   from_hdr.offset);
+   params[param] |= mc_enc(32, 32,
+   cfg->extracts[i].extract.
+   from_hdr.field);
+   

[dpdk-dev] [PATCHv6 06/33] bus/fslmc: add mc dpio object support

[Hemant Agrawal]

This patch adds the DPIO object support in MC driver.

DPIO - Data Path Input Output represent the processing
context to access the QBMAN HW for packet I/O.

Signed-off-by: Alex Marginean 
Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile  |   1 +
 drivers/bus/fslmc/mc/dpio.c | 279 +++
 drivers/bus/fslmc/mc/fsl_dpio.h | 282 
 drivers/bus/fslmc/mc/fsl_dpio_cmd.h | 121 
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   8 +
 5 files changed, 691 insertions(+)
 create mode 100644 drivers/bus/fslmc/mc/dpio.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpio.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpio_cmd.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 15ab89a..53ab17b 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -50,6 +50,7 @@ LIBABIVER := 1
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/dpni.c \
+mc/dpio.c \
 mc/mc_sys.c
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
diff --git a/drivers/bus/fslmc/mc/dpio.c b/drivers/bus/fslmc/mc/dpio.c
new file mode 100644
index 000..7f7359d
--- /dev/null
+++ b/drivers/bus/fslmc/mc/dpio.c
@@ -0,0 +1,279 @@
+/*-
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ *   BSD LICENSE
+ *
+ * Copyright 2013-2016 Freescale Semiconductor Inc.
+ * Copyright (c) 2016 NXP.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *   GPL LICENSE SUMMARY
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#include 
+#include 
+#include 
+#include 
+
+int dpio_open(struct fsl_mc_io *mc_io,
+ uint32_t cmd_flags,
+ int dpio_id,
+ uint16_t *token)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPIO_CMDID_OPEN,
+ cmd_flags,
+ 0);
+   DPIO_CMD_OPEN(cmd, dpio_id);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   *token = MC_CMD_HDR_READ_TOKEN(cmd.header);
+
+   return 0;
+}
+
+int dpio_close(struct fsl_mc_io *mc_io,
+  uint32_t cmd_flags,
+  uint16_t token)
+{
+   struct mc_command cmd = { 0 };
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPIO_CMDID_CLOSE,
+ cmd_flags,
+ token);
+
+   /* send command to mc*/
+   return mc_send_command(mc_io, &cmd);
+}
+
+int dpio_create(struct fsl_mc_io   *mc_io,
+   uint16_tdprc_token,
+   uint32_tcmd_flags,
+   const struct dpio_cfg   *cfg,
+   uint32_t*obj_id)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPIO_CMDID_CREATE,
+ cmd_flags,
+ dprc_token);
+   DPIO_CMD_CREATE(cmd, cfg);
+

[dpdk-dev] [PATCHv6 07/33] bus/fslmc: add mc dpbp object support

[Hemant Agrawal]

DPBP object represent a hw based buffer pool instance
in the DPAA2 hardware.

Signed-off-by: Alex Marginean 
Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile  |   1 +
 drivers/bus/fslmc/mc/dpbp.c | 237 
 drivers/bus/fslmc/mc/fsl_dpbp.h | 227 ++
 drivers/bus/fslmc/mc/fsl_dpbp_cmd.h |  83 ++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   5 +
 5 files changed, 553 insertions(+)
 create mode 100644 drivers/bus/fslmc/mc/dpbp.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpbp.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpbp_cmd.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 53ab17b..628e517 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -50,6 +50,7 @@ LIBABIVER := 1
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/dpni.c \
+mc/dpbp.c \
 mc/dpio.c \
 mc/mc_sys.c
 
diff --git a/drivers/bus/fslmc/mc/dpbp.c b/drivers/bus/fslmc/mc/dpbp.c
new file mode 100644
index 000..0d93e0c
--- /dev/null
+++ b/drivers/bus/fslmc/mc/dpbp.c
@@ -0,0 +1,237 @@
+/*-
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ *   BSD LICENSE
+ *
+ * Copyright 2013-2016 Freescale Semiconductor Inc.
+ * Copyright (c) 2016 NXP.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *   GPL LICENSE SUMMARY
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#include 
+#include 
+#include 
+#include 
+
+int dpbp_open(struct fsl_mc_io *mc_io,
+ uint32_t cmd_flags,
+ int dpbp_id,
+ uint16_t *token)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPBP_CMDID_OPEN,
+ cmd_flags,
+ 0);
+   DPBP_CMD_OPEN(cmd, dpbp_id);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   *token = MC_CMD_HDR_READ_TOKEN(cmd.header);
+
+   return err;
+}
+
+int dpbp_close(struct fsl_mc_io *mc_io,
+  uint32_t cmd_flags,
+  uint16_t token)
+{
+   struct mc_command cmd = { 0 };
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPBP_CMDID_CLOSE, cmd_flags,
+ token);
+
+   /* send command to mc*/
+   return mc_send_command(mc_io, &cmd);
+}
+
+int dpbp_create(struct fsl_mc_io *mc_io,
+   uint16_t dprc_token,
+   uint32_t cmd_flags,
+   const struct dpbp_cfg *cfg,
+   uint32_t *obj_id)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   (void)(cfg); /* unused */
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPBP_CMDID_CREATE,
+ cmd_flags,
+ dprc_token);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   CMD_CREA

[dpdk-dev] [PATCHv6 08/33] bus/fslmc: add mc dpseci object support

[Hemant Agrawal]

dpseci represent a instance of SEC HW in DPAA2.

Signed-off-by: Cristian Sovaiala 
Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile  |   1 +
 drivers/bus/fslmc/mc/dpseci.c   | 534 ++
 drivers/bus/fslmc/mc/fsl_dpseci.h   | 668 
 drivers/bus/fslmc/mc/fsl_dpseci_cmd.h   | 255 +++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  10 +
 5 files changed, 1468 insertions(+)
 create mode 100644 drivers/bus/fslmc/mc/dpseci.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpseci.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpseci_cmd.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 628e517..4a118a3 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -50,6 +50,7 @@ LIBABIVER := 1
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/dpni.c \
+mc/dpseci.c \
 mc/dpbp.c \
 mc/dpio.c \
 mc/mc_sys.c
diff --git a/drivers/bus/fslmc/mc/dpseci.c b/drivers/bus/fslmc/mc/dpseci.c
new file mode 100644
index 000..bc8c9c5
--- /dev/null
+++ b/drivers/bus/fslmc/mc/dpseci.c
@@ -0,0 +1,534 @@
+/*-
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ *   BSD LICENSE
+ *
+ * Copyright 2013-2016 Freescale Semiconductor Inc.
+ * Copyright (c) 2016 NXP.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *   GPL LICENSE SUMMARY
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#include 
+#include 
+#include 
+#include 
+
+int dpseci_open(struct fsl_mc_io *mc_io,
+   uint32_t cmd_flags,
+   int dpseci_id,
+   uint16_t *token)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPSECI_CMDID_OPEN,
+ cmd_flags,
+ 0);
+   DPSECI_CMD_OPEN(cmd, dpseci_id);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   *token = MC_CMD_HDR_READ_TOKEN(cmd.header);
+
+   return 0;
+}
+
+int dpseci_close(struct fsl_mc_io *mc_io,
+uint32_t cmd_flags,
+uint16_t token)
+{
+   struct mc_command cmd = { 0 };
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPSECI_CMDID_CLOSE,
+ cmd_flags,
+ token);
+
+   /* send command to mc*/
+   return mc_send_command(mc_io, &cmd);
+}
+
+int dpseci_create(struct fsl_mc_io *mc_io,
+ uint16_t  dprc_token,
+ uint32_t  cmd_flags,
+ const struct dpseci_cfg   *cfg,
+ uint32_t  *obj_id)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPSECI_CMDID_CREATE,
+ cmd_flags,
+ dprc_token);
+   DPSECI_CMD_CREATE(cmd, cfg);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd

[dpdk-dev] [PATCHv6 09/33] eal/vfio: adding vfio utility functions in map file

[Hemant Agrawal]

adding extra vfio utility functions to map file.
They will be used by other vfio supported buses like fslmc bus
for NXP DPAA2 devices

Signed-off-by: Hemant Agrawal 
---
 lib/librte_eal/bsdapp/eal/rte_eal_version.map   | 3 +++
 lib/librte_eal/linuxapp/eal/rte_eal_version.map | 3 +++
 2 files changed, 6 insertions(+)

diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map 
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 2cf1ac8..a74625a 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -183,5 +183,8 @@ DPDK_17.02 {
rte_bus_register;
rte_bus_scan;
rte_bus_unregister;
+   vfio_get_container_fd;
+   vfio_get_group_fd;
+   vfio_get_group_no;
 
 } DPDK_16.11;
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map 
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index 3c68ff5..99d4446 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -187,5 +187,8 @@ DPDK_17.02 {
rte_bus_register;
rte_bus_scan;
rte_bus_unregister;
+   vfio_get_container_fd;
+   vfio_get_group_fd;
+   vfio_get_group_no;
 
 } DPDK_16.11;
-- 
1.9.1

[dpdk-dev] [PATCHv6 10/33] bus/fslmc: add vfio support

[Hemant Agrawal]

Add support for using VFIO for dpaa2 based fsl-mc bus.

There are some differences in the way vfio used for fsl-mc bus
from the eal vfio.
 - The scanning of bus for individual objects on the basis of
   the DPRC container.
 - The use and mapping of MC portal for object access

With the evolution of bus model, they canbe further aligned with
eal vfio code.

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile  |   2 +
 drivers/bus/fslmc/fslmc_bus.c   |  10 +
 drivers/bus/fslmc/fslmc_vfio.c  | 450 
 drivers/bus/fslmc/fslmc_vfio.h  |  74 +
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   1 +
 5 files changed, 537 insertions(+)
 create mode 100644 drivers/bus/fslmc/fslmc_vfio.c
 create mode 100644 drivers/bus/fslmc/fslmc_vfio.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 4a118a3..70e 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -41,6 +41,7 @@ CFLAGS += "-Wno-strict-aliasing"
 
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/mc
+CFLAGS += -I$(RTE_SDK)/lib/librte_eal/linuxapp/eal
 
 # versioning export map
 EXPORT_MAP := rte_bus_fslmc_version.map
@@ -55,6 +56,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/dpio.c \
 mc/mc_sys.c
 
+SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_vfio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
 
 # library dependencies
diff --git a/drivers/bus/fslmc/fslmc_bus.c b/drivers/bus/fslmc/fslmc_bus.c
index 8a4f519..ee794e7 100644
--- a/drivers/bus/fslmc/fslmc_bus.c
+++ b/drivers/bus/fslmc/fslmc_bus.c
@@ -42,6 +42,7 @@
 #include 
 
 #include "rte_fslmc.h"
+#include "fslmc_vfio.h"
 
 #define FSLMC_BUS_LOG(level, fmt, args...) \
RTE_LOG(level, EAL, "%s(): " fmt "\n", __func__, ##args)
@@ -51,6 +52,15 @@
 static int
 rte_fslmc_scan(void)
 {
+   if (fslmc_vfio_setup_group()) {
+   FSLMC_BUS_LOG(ERR, "fslmc: Unable to setup VFIO");
+   return -1;
+   }
+   if (fslmc_vfio_process_group()) {
+   FSLMC_BUS_LOG(ERR, "fslmc: Unable to setup devices");
+   return -1;
+   }
+   RTE_LOG(INFO, EAL, "fslmc: Bus scan completed\n");
return 0;
 }
 
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
new file mode 100644
index 000..73db595
--- /dev/null
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -0,0 +1,450 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright (c) 2015-2016 Freescale Semiconductor, Inc. All rights reserved.
+ *   Copyright (c) 2016 NXP. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Freescale Semiconductor, Inc nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include "rte_fslmc.h"
+#include "fslmc_vfio.h"
+
+#define VFIO_MAX_CONTAINERS1
+
+#define FSLMC_VFIO_LOG(level, fmt, args...) \
+   RTE_LOG(level, EAL, "%s(): " fmt "\n", __func__, ##args)
+
+/** Pathname of FSL-MC devices directory. */
+#define SYSFS_FSL_MC_DEVICES "/sys/bus/fsl-mc/devices"
+
+/* Number of VFIO containers & groups with in */
+static struct fslmc_vfio_group vfio_groups[VFIO_MAX_GRP];
+static struct fslmc_vfio_container vfio_containers[V

[dpdk-dev] [PATCHv6 11/33] bus/fslmc: scan for net and sec devices

[Hemant Agrawal]

This patch will add support in fslmc vfio process to
scan and parse the dpni and dpseci object for net and crypto
devices. It will add the scanned devices to the fslmc bus.

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/fslmc_vfio.c | 63 +-
 1 file changed, 62 insertions(+), 1 deletion(-)

diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 73db595..0d4c0a2 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -210,6 +210,48 @@ static int64_t vfio_map_mcp_obj(struct fslmc_vfio_group 
*group, char *mcp_obj)
return v_addr;
 }
 
+static inline int
+dpaa2_compare_dpaa2_dev(const struct rte_dpaa2_device *dev,
+const struct rte_dpaa2_device *dev2)
+{
+   /*not the same family device */
+   if (dev->dev_type != DPAA2_MC_DPNI_DEVID ||
+   dev->dev_type != DPAA2_MC_DPSECI_DEVID)
+   return -1;
+
+   if (dev->object_id == dev2->object_id)
+   return 0;
+   else
+   return 1;
+}
+
+static void
+fslmc_bus_add_device(struct rte_dpaa2_device *dev)
+{
+   struct rte_fslmc_device_list *dev_l;
+
+   dev_l = &rte_fslmc_bus.device_list;
+
+   /* device is valid, add in list (sorted) */
+   if (TAILQ_EMPTY(dev_l)) {
+   TAILQ_INSERT_TAIL(dev_l, dev, next);
+   } else {
+   struct rte_dpaa2_device *dev2;
+   int ret;
+
+   TAILQ_FOREACH(dev2, dev_l, next) {
+   ret = dpaa2_compare_dpaa2_dev(dev, dev2);
+   if (ret <= 0)
+   continue;
+
+   TAILQ_INSERT_BEFORE(dev2, dev, next);
+   return;
+   }
+
+   TAILQ_INSERT_TAIL(dev_l, dev, next);
+   }
+}
+
 /* Following function shall fetch total available list of MC devices
  * from VFIO container & populate private list of devices and other
  * data structures
@@ -218,7 +260,7 @@ int fslmc_vfio_process_group(void)
 {
struct fslmc_vfio_device *vdev;
struct vfio_device_info device_info = { .argsz = sizeof(device_info) };
-   char *temp_obj, *object_type __rte_unused, *mcp_obj, *dev_name;
+   char *temp_obj, *object_type, *mcp_obj, *dev_name;
int32_t object_id, i, dev_fd;
DIR *d;
struct dirent *dir;
@@ -348,6 +390,25 @@ int fslmc_vfio_process_group(void)
FSLMC_VFIO_LOG(ERR, "DPAA2 VFIO_DEVICE_GET_INFO fail");
goto FAILURE;
}
+   if (!strcmp(object_type, "dpni") ||
+   !strcmp(object_type, "dpseci")) {
+   struct rte_dpaa2_device *dev;
+
+   dev = malloc(sizeof(struct rte_dpaa2_device));
+   if (dev == NULL)
+   return -1;
+
+   memset(dev, 0, sizeof(*dev));
+   /* store hw_id of dpni/dpseci device */
+   dev->object_id = object_id;
+   dev->dev_type = (strcmp(object_type, "dpseci")) ?
+   DPAA2_MC_DPNI_DEVID : DPAA2_MC_DPSECI_DEVID;
+
+   FSLMC_VFIO_LOG(DEBUG, "DPAA2: Added [%s-%d]\n",
+ object_type, object_id);
+
+   fslmc_bus_add_device(dev);
+   }
}
closedir(d);
 
-- 
1.9.1

[dpdk-dev] [PATCHv6 13/33] doc: add dpaa2 nic details

[Hemant Agrawal]

This patch adds the NXP dpaa2 architecture and pmd details
in the Network interfaces section.

Signed-off-by: Hemant Agrawal 
Acked-by: John McNamara 
---
 MAINTAINERS|   1 +
 doc/guides/nics/dpaa2.rst  | 593 +
 doc/guides/nics/features/dpaa2.ini |   9 +
 doc/guides/nics/index.rst  |   1 +
 doc/guides/rel_notes/release_17_02.rst |  12 +-
 5 files changed, 615 insertions(+), 1 deletion(-)
 create mode 100644 doc/guides/nics/dpaa2.rst
 create mode 100644 doc/guides/nics/features/dpaa2.ini

diff --git a/MAINTAINERS b/MAINTAINERS
index 7e274e0..fb85351 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -359,6 +359,7 @@ M: Hemant Agrawal 
 F: drivers/bus/fslmc/
 F: drivers/common/dpaa2/
 F: drivers/net/dpaa2/
+F: doc/guides/nics/dpaa2.rst
 
 QLogic bnx2x
 M: Harish Patil 
diff --git a/doc/guides/nics/dpaa2.rst b/doc/guides/nics/dpaa2.rst
new file mode 100644
index 000..f0d7a26
--- /dev/null
+++ b/doc/guides/nics/dpaa2.rst
@@ -0,0 +1,593 @@
+..  BSD LICENSE
+Copyright (C) NXP. 2016.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+* Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+* Redistributions in binary form must reproduce the above copyright
+notice, this list of conditions and the following disclaimer in
+the documentation and/or other materials provided with the
+distribution.
+* Neither the name of NXP nor the names of its
+contributors may be used to endorse or promote products derived
+from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+DPAA2 Poll Mode Driver
+==
+
+The DPAA2 NIC PMD (**librte_pmd_dpaa2**) provides poll mode driver
+support for the inbuilt NIC found in the **NXP DPAA2** SoC family.
+
+More information can be found at `NXP Official Website
+`_.
+
+NXP DPAA2 (Data Path Acceleration Architecture Gen2)
+
+
+This section provides an overview of the NXP DPAA2 architecture
+and how it is integrated into the DPDK.
+
+Contents summary
+
+- DPAA2 overview
+- Overview of DPAA2 objects
+- DPAA2 driver architecture overview
+
+DPAA2 Overview
+~~
+
+Reference: `FSL MC BUS in Linux Kernel 
`_.
+
+DPAA2 is a hardware architecture designed for high-speed network
+packet processing.  DPAA2 consists of sophisticated mechanisms for
+processing Ethernet packets, queue management, buffer management,
+autonomous L2 switching, virtual Ethernet bridging, and accelerator
+(e.g. crypto) sharing.
+
+A DPAA2 hardware component called the Management Complex (or MC) manages the
+DPAA2 hardware resources.  The MC provides an object-based abstraction for
+software drivers to use the DPAA2 hardware.
+
+The MC uses DPAA2 hardware resources such as queues, buffer pools, and
+network ports to create functional objects/devices such as network
+interfaces, an L2 switch, or accelerator instances.
+
+The MC provides memory-mapped I/O command interfaces (MC portals)
+which DPAA2 software drivers use to operate on DPAA2 objects:
+
+The diagram below shows an overview of the DPAA2 resource management
+architecture:
+
+.. code-block:: console
+
+  +--+
+  |  OS  |
+  |DPAA2 drivers |
+  | ||
+  +-|+
+|
+| (create,discover,connect
+|  config,use,destroy)
+|
+  DPAA2 |
+  +| mc portal |-+
+  | ||
+  |   +- - - - - - - - - - - - -V- - -+  |
+  |   |

[dpdk-dev] [PATCHv6 12/33] net/dpaa2: introducing NXP dpaa2 pmd driver

[Hemant Agrawal]

add support for fsl-mc bus based dpaa2 pmd driver.

Signed-off-by: Hemant Agrawal 
---
 MAINTAINERS |   1 +
 config/common_base  |   4 +
 config/defconfig_arm64-dpaa2-linuxapp-gcc   |   5 +
 drivers/bus/Makefile|   2 +
 drivers/bus/fslmc/Makefile  |   4 +
 drivers/common/Makefile |   4 +
 drivers/common/dpaa2/Makefile   |   4 +
 drivers/common/dpaa2/qbman/Makefile |   4 +
 drivers/net/Makefile|   2 +-
 drivers/net/dpaa2/Makefile  |  59 
 drivers/net/dpaa2/dpaa2_ethdev.c| 142 
 drivers/net/dpaa2/dpaa2_ethdev.h|  44 +
 drivers/net/dpaa2/rte_pmd_dpaa2_version.map |   4 +
 mk/rte.app.mk   |   3 +
 14 files changed, 281 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/dpaa2/Makefile
 create mode 100644 drivers/net/dpaa2/dpaa2_ethdev.c
 create mode 100644 drivers/net/dpaa2/dpaa2_ethdev.h
 create mode 100644 drivers/net/dpaa2/rte_pmd_dpaa2_version.map

diff --git a/MAINTAINERS b/MAINTAINERS
index 4a13140..7e274e0 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -358,6 +358,7 @@ NXP dpaa2
 M: Hemant Agrawal 
 F: drivers/bus/fslmc/
 F: drivers/common/dpaa2/
+F: drivers/net/dpaa2/
 
 QLogic bnx2x
 M: Harish Patil 
diff --git a/config/common_base b/config/common_base
index bf1de8f..7054999 100644
--- a/config/common_base
+++ b/config/common_base
@@ -292,6 +292,10 @@ CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_MBOX=n
 #
 CONFIG_RTE_LIBRTE_FSLMC_BUS=n
 
+#
+# Compile burst-oriented NXP DPAA2 PMD driver
+#
+CONFIG_RTE_LIBRTE_DPAA2_PMD=n
 # Compile burst-oriented VIRTIO PMD driver
 #
 CONFIG_RTE_LIBRTE_VIRTIO_PMD=y
diff --git a/config/defconfig_arm64-dpaa2-linuxapp-gcc 
b/config/defconfig_arm64-dpaa2-linuxapp-gcc
index 365ae5a..e63ff56 100644
--- a/config/defconfig_arm64-dpaa2-linuxapp-gcc
+++ b/config/defconfig_arm64-dpaa2-linuxapp-gcc
@@ -46,3 +46,8 @@ CONFIG_RTE_MAX_NUMA_NODES=1
 # Compile NXP DPAA2 FSL-MC Bus
 #
 CONFIG_RTE_LIBRTE_FSLMC_BUS=y
+
+#
+# Compile burst-oriented NXP DPAA2 PMD driver
+#
+CONFIG_RTE_LIBRTE_DPAA2_PMD=y
diff --git a/drivers/bus/Makefile b/drivers/bus/Makefile
index 60e9764..8f7864b 100644
--- a/drivers/bus/Makefile
+++ b/drivers/bus/Makefile
@@ -31,6 +31,8 @@
 
 include $(RTE_SDK)/mk/rte.vars.mk
 
+CONFIG_RTE_LIBRTE_FSLMC_BUS = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
+
 DIRS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc
 
 include $(RTE_SDK)/mk/rte.subdir.mk
diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 70e..ce799da 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -35,6 +35,10 @@ include $(RTE_SDK)/mk/rte.vars.mk
 #
 LIB = librte_bus_fslmc.a
 
+ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_PMD),y)
+CONFIG_RTE_LIBRTE_FSLMC_BUS = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
+endif
+
 CFLAGS += -O3
 CFLAGS += $(WERROR_FLAGS)
 CFLAGS += "-Wno-strict-aliasing"
diff --git a/drivers/common/Makefile b/drivers/common/Makefile
index cba1134..b52931c 100644
--- a/drivers/common/Makefile
+++ b/drivers/common/Makefile
@@ -31,6 +31,10 @@
 
 include $(RTE_SDK)/mk/rte.vars.mk
 
+ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_PMD),y)
+CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
+endif
+
 ifeq ($(CONFIG_RTE_LIBRTE_FSLMC_BUS),y)
 CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_FSLMC_BUS)
 endif
diff --git a/drivers/common/dpaa2/Makefile b/drivers/common/dpaa2/Makefile
index 9681729..87f08bb 100644
--- a/drivers/common/dpaa2/Makefile
+++ b/drivers/common/dpaa2/Makefile
@@ -31,6 +31,10 @@
 
 include $(RTE_SDK)/mk/rte.vars.mk
 
+ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_PMD),y)
+CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
+endif
+
 ifeq ($(CONFIG_RTE_LIBRTE_FSLMC_BUS),y)
 CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_FSLMC_BUS)
 endif
diff --git a/drivers/common/dpaa2/qbman/Makefile 
b/drivers/common/dpaa2/qbman/Makefile
index 7ac1ba7..18bca6b 100644
--- a/drivers/common/dpaa2/qbman/Makefile
+++ b/drivers/common/dpaa2/qbman/Makefile
@@ -36,6 +36,10 @@ include $(RTE_SDK)/mk/rte.vars.mk
 #
 LIB = librte_common_dpaa2_qbman.a
 
+ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_PMD),y)
+CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
+endif
+
 ifeq ($(CONFIG_RTE_LIBRTE_FSLMC_BUS),y)
 CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_FSLMC_BUS)
 endif
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 40fc333..c2f64ce 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -35,6 +35,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_PMD_AF_PACKET) += af_packet
 DIRS-$(CONFIG_RTE_LIBRTE_BNX2X_PMD) += bnx2x
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_BOND) += bonding
 DIRS-$(CONFIG_RTE_LIBRTE_CXGBE_PMD) += cxgbe
+DIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2
 DIRS-$(CONFIG_RTE_LIBRTE_E1000_PMD) += e1000
 DIRS-$(CONFIG_RTE_LIBRTE_ENA_PMD) += ena
 DIRS-$(CONFIG_RTE_LIBRTE_ENIC_PMD) += enic
@@ -57,7 +58,6 @@ DIRS-$(CONFIG

[dpdk-dev] [PATCHv6 15/33] drivers/common/dpaa2: dpio portal driver

[Hemant Agrawal]

The portal driver is bound to DPIO objects discovered on the fsl-mc bus and
provides services that:
- allow other drivers, such as the Ethernet driver, to enqueue and dequeue
  frames for their respective objects

A system will typically allocate 1 DPIO object per CPU to allow queuing
operations to happen simultaneously across all CPUs.

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/Makefile  |   3 +
 drivers/bus/fslmc/fslmc_vfio.c  |  17 +-
 drivers/bus/fslmc/fslmc_vfio.h  |   5 +
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c| 364 
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h|  60 +
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h |  68 ++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   2 +
 7 files changed, 518 insertions(+), 1 deletion(-)
 create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
 create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
 create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 77e0a1b..311e0e9 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -50,6 +50,7 @@ CFLAGS += "-Wno-strict-aliasing"
 
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/mc
+CFLAGS += -I$(RTE_SDK)/drivers/common/dpaa2/qbman/include
 CFLAGS += -I$(RTE_SDK)/lib/librte_eal/linuxapp/eal
 
 # versioning export map
@@ -65,10 +66,12 @@ SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/dpio.c \
 mc/mc_sys.c
 
+SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_vfio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
 
 # library dependencies
 DEPDIRS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += lib/librte_eal
+DEPDIRS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += lib/librte_common_dpaa2_qbman
 
 include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 0d4c0a2..2d7bcd9 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -61,6 +61,9 @@
 #include "rte_fslmc.h"
 #include "fslmc_vfio.h"
 
+#include "portal/dpaa2_hw_pvt.h"
+#include "portal/dpaa2_hw_dpio.h"
+
 #define VFIO_MAX_CONTAINERS1
 
 #define FSLMC_VFIO_LOG(level, fmt, args...) \
@@ -261,12 +264,13 @@ int fslmc_vfio_process_group(void)
struct fslmc_vfio_device *vdev;
struct vfio_device_info device_info = { .argsz = sizeof(device_info) };
char *temp_obj, *object_type, *mcp_obj, *dev_name;
-   int32_t object_id, i, dev_fd;
+   int32_t object_id, i, dev_fd, ret;
DIR *d;
struct dirent *dir;
char path[PATH_MAX];
int64_t v_addr;
int ndev_count;
+   int dpio_count = 0;
struct fslmc_vfio_group *group = &vfio_groups[0];
static int process_once;
 
@@ -409,9 +413,20 @@ int fslmc_vfio_process_group(void)
 
fslmc_bus_add_device(dev);
}
+   if (!strcmp(object_type, "dpio")) {
+   ret = dpaa2_create_dpio_device(vdev,
+  &device_info,
+  object_id);
+   if (!ret)
+   dpio_count++;
+   }
}
closedir(d);
 
+   ret = dpaa2_affine_qbman_swp();
+   if (ret)
+   FSLMC_VFIO_LOG(DEBUG, "Error in affining qbman swp %d", ret);
+
return 0;
 
 FAILURE:
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index 5e58211..39994dd 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -71,4 +71,9 @@ int vfio_dmamap_mem_region(
 int fslmc_vfio_setup_group(void);
 int fslmc_vfio_process_group(void);
 
+/* create dpio device */
+int dpaa2_create_dpio_device(struct fslmc_vfio_device *vdev,
+struct vfio_device_info *obj_info,
+int object_id);
+
 #endif /* _FSLMC_VFIO_H_ */
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
new file mode 100644
index 000..dd6de4c
--- /dev/null
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -0,0 +1,364 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright (c) 2016 Freescale Semiconductor, Inc. All rights reserved.
+ *   Copyright (c) 2016 NXP. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the

[dpdk-dev] [PATCHv6 14/33] bus/fslmc: add debug log message support

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 config/common_base|  7 +++
 config/defconfig_arm64-dpaa2-linuxapp-gcc |  5 ++
 drivers/bus/fslmc/Makefile|  5 ++
 drivers/bus/fslmc/fslmc_logs.h| 76 +++
 drivers/common/dpaa2/qbman/Makefile   |  5 ++
 drivers/net/dpaa2/Makefile|  5 ++
 drivers/net/dpaa2/dpaa2_ethdev.c  |  9 +++-
 7 files changed, 110 insertions(+), 2 deletions(-)
 create mode 100644 drivers/bus/fslmc/fslmc_logs.h

diff --git a/config/common_base b/config/common_base
index 7054999..d1bf7a0 100644
--- a/config/common_base
+++ b/config/common_base
@@ -296,6 +296,13 @@ CONFIG_RTE_LIBRTE_FSLMC_BUS=n
 # Compile burst-oriented NXP DPAA2 PMD driver
 #
 CONFIG_RTE_LIBRTE_DPAA2_PMD=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_INIT=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_DRIVER=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_RX=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_TX=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_TX_FREE=n
+
+#
 # Compile burst-oriented VIRTIO PMD driver
 #
 CONFIG_RTE_LIBRTE_VIRTIO_PMD=y
diff --git a/config/defconfig_arm64-dpaa2-linuxapp-gcc 
b/config/defconfig_arm64-dpaa2-linuxapp-gcc
index e63ff56..eb12511 100644
--- a/config/defconfig_arm64-dpaa2-linuxapp-gcc
+++ b/config/defconfig_arm64-dpaa2-linuxapp-gcc
@@ -51,3 +51,8 @@ CONFIG_RTE_LIBRTE_FSLMC_BUS=y
 # Compile burst-oriented NXP DPAA2 PMD driver
 #
 CONFIG_RTE_LIBRTE_DPAA2_PMD=y
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_INIT=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_DRIVER=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_RX=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_TX=n
+CONFIG_RTE_LIBRTE_DPAA2_DEBUG_TX_FREE=n
diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index ce799da..77e0a1b 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -39,8 +39,13 @@ ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_PMD),y)
 CONFIG_RTE_LIBRTE_FSLMC_BUS = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
 endif
 
+ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_DEBUG_INIT),y)
+CFLAGS += -O0 -g
+CFLAGS += "-Wno-error"
+else
 CFLAGS += -O3
 CFLAGS += $(WERROR_FLAGS)
+endif
 CFLAGS += "-Wno-strict-aliasing"
 
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
diff --git a/drivers/bus/fslmc/fslmc_logs.h b/drivers/bus/fslmc/fslmc_logs.h
new file mode 100644
index 000..a890e6c
--- /dev/null
+++ b/drivers/bus/fslmc/fslmc_logs.h
@@ -0,0 +1,76 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright (c) 2016 NXP. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of NXP nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef _FSLMC_LOGS_H_
+#define _FSLMC_LOGS_H_
+
+#define PMD_INIT_LOG(level, fmt, args...) \
+   RTE_LOG(level, PMD, "%s(): " fmt "\n", __func__, ##args)
+
+#ifdef RTE_LIBRTE_DPAA2_DEBUG_INIT
+#define PMD_INIT_FUNC_TRACE() PMD_INIT_LOG(DEBUG, " >>")
+#else
+#define PMD_INIT_FUNC_TRACE() do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_DPAA2_DEBUG_RX
+#define PMD_RX_LOG(level, fmt, args...) \
+   RTE_LOG(level, PMD, "%s(): " fmt "\n", __func__, ## args)
+#else
+#define PMD_RX_LOG(level, fmt, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_DPAA2_DEBUG_TX
+#define PMD_TX_LOG(level, fmt, args...) \
+   RTE_LOG(level, PMD, "%s(): " fmt "\n", __func__, ## args)
+#else
+#define PMD_TX_LOG(level, fmt, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_DPAA2_DEBUG_TX_FREE
+#define PMD_TX_FREE_LOG(level, fmt, args...) \
+   RTE_LOG(level, PMD, "%s(): " fmt "\n", __func__, ## args)
+#else
+#define PMD_TX_FREE_LOG(level, fmt, args...) do { } while (0)
+#endif
+
+#ifdef RTE_LIBRTE_D

[dpdk-dev] [PATCHv6 16/33] drivers/pool/dpaa2: adding hw offloaded mempool

[Hemant Agrawal]

Adding NXP DPAA2 architecture specific mempool support
Each mempool instance is represented by a DPBP object
from the FSL-MC bus.

This patch also registers a dpaa2 type MEMPOOL OPS

Signed-off-by: Hemant Agrawal 
---
 MAINTAINERS   |   1 +
 config/common_base|   5 +
 config/defconfig_arm64-dpaa2-linuxapp-gcc |   8 +
 drivers/Makefile  |   1 +
 drivers/bus/fslmc/Makefile|   2 +
 drivers/bus/fslmc/fslmc_vfio.c|   9 +-
 drivers/bus/fslmc/fslmc_vfio.h|   2 +
 drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c  | 137 +++
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h   |  20 ++
 drivers/bus/fslmc/rte_bus_fslmc_version.map   |   2 +
 drivers/common/Makefile   |   6 +-
 drivers/common/dpaa2/Makefile |   4 +
 drivers/common/dpaa2/qbman/Makefile   |   4 +
 drivers/pool/Makefile |  40 +++
 drivers/pool/dpaa2/Makefile   |  71 ++
 drivers/pool/dpaa2/dpaa2_hw_mempool.c | 339 ++
 drivers/pool/dpaa2/dpaa2_hw_mempool.h |  95 
 drivers/pool/dpaa2/rte_pool_dpaa2_version.map |   8 +
 mk/rte.app.mk |   1 +
 19 files changed, 753 insertions(+), 2 deletions(-)
 create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c
 create mode 100644 drivers/pool/Makefile
 create mode 100644 drivers/pool/dpaa2/Makefile
 create mode 100644 drivers/pool/dpaa2/dpaa2_hw_mempool.c
 create mode 100644 drivers/pool/dpaa2/dpaa2_hw_mempool.h
 create mode 100644 drivers/pool/dpaa2/rte_pool_dpaa2_version.map

diff --git a/MAINTAINERS b/MAINTAINERS
index fb85351..a3ebcd5 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -359,6 +359,7 @@ M: Hemant Agrawal 
 F: drivers/bus/fslmc/
 F: drivers/common/dpaa2/
 F: drivers/net/dpaa2/
+F: drivers/pool/dpaa2/
 F: doc/guides/nics/dpaa2.rst
 
 QLogic bnx2x
diff --git a/config/common_base b/config/common_base
index d1bf7a0..dd3de11 100644
--- a/config/common_base
+++ b/config/common_base
@@ -288,6 +288,11 @@ CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_DRIVER=n
 CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_MBOX=n
 
 #
+# Compile Support Libraries for NXP DPAA2
+#
+CONFIG_RTE_LIBRTE_DPAA2_POOL=n
+
+#
 # Compile NXP DPAA2 FSL-MC Bus
 #
 CONFIG_RTE_LIBRTE_FSLMC_BUS=n
diff --git a/config/defconfig_arm64-dpaa2-linuxapp-gcc 
b/config/defconfig_arm64-dpaa2-linuxapp-gcc
index eb12511..3cdb31b 100644
--- a/config/defconfig_arm64-dpaa2-linuxapp-gcc
+++ b/config/defconfig_arm64-dpaa2-linuxapp-gcc
@@ -42,6 +42,14 @@ CONFIG_RTE_ARCH_ARM_TUNE="cortex-a57+fp+simd"
 CONFIG_RTE_MAX_LCORE=8
 CONFIG_RTE_MAX_NUMA_NODES=1
 
+CONFIG_RTE_PKTMBUF_HEADROOM=256
+
+#
+# Compile Support Libraries for DPAA2
+#
+CONFIG_RTE_LIBRTE_DPAA2_POOL=n
+CONFIG_RTE_MBUF_DEFAULT_MEMPOOL_OPS="dpaa2"
+
 #
 # Compile NXP DPAA2 FSL-MC Bus
 #
diff --git a/drivers/Makefile b/drivers/Makefile
index bdae63b..9fd268e 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -33,6 +33,7 @@ include $(RTE_SDK)/mk/rte.vars.mk
 
 DIRS-y += common
 DIRS-y += bus
+DIRS-y += pool
 DIRS-y += net
 DIRS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += crypto
 
diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 311e0e9..263c4fd 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -51,6 +51,7 @@ CFLAGS += "-Wno-strict-aliasing"
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/mc
 CFLAGS += -I$(RTE_SDK)/drivers/common/dpaa2/qbman/include
+CFLAGS += -I$(RTE_SDK)/drivers/pool/dpaa2
 CFLAGS += -I$(RTE_SDK)/lib/librte_eal/linuxapp/eal
 
 # versioning export map
@@ -67,6 +68,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/mc_sys.c
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpio.c
+SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpbp.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_vfio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
 
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 2d7bcd9..fc017fc 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -270,7 +270,7 @@ int fslmc_vfio_process_group(void)
char path[PATH_MAX];
int64_t v_addr;
int ndev_count;
-   int dpio_count = 0;
+   int dpio_count = 0, dpbp_count = 0;
struct fslmc_vfio_group *group = &vfio_groups[0];
static int process_once;
 
@@ -420,6 +420,11 @@ int fslmc_vfio_process_group(void)
if (!ret)
dpio_count++;
}
+   if (!strcmp(object_type, "dpbp")) {
+   ret = dpaa2_create_dpbp_device(object_id);
+   if (!ret)
+   dpbp_count++;
+   }
}
closedir(d);
 
@@ -427,6 +432,8 @@ int fslmc_vfio_process_group(void)
if (ret)
FSLM

[dpdk-dev] [PATCHv6 17/33] drivers/common/dpaa2: dpio routine to affine to crypto threads

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c| 45 +
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h|  3 ++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  1 +
 3 files changed, 49 insertions(+)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
index dd6de4c..bd1f643 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -276,6 +276,51 @@ static inline struct dpaa2_dpio_dev 
*dpaa2_get_qbman_swp(void)
 }
 
 int
+dpaa2_affine_qbman_swp_sec(void)
+{
+   unsigned int lcore_id = rte_lcore_id();
+   uint64_t tid = syscall(SYS_gettid);
+
+   if (lcore_id == LCORE_ID_ANY)
+   lcore_id = rte_get_master_lcore();
+   /* if the core id is not supported */
+   else if (lcore_id >= RTE_MAX_LCORE)
+   return -1;
+
+   if (dpaa2_io_portal[lcore_id].sec_dpio_dev) {
+   PMD_DRV_LOG(INFO, "DPAA Portal=0x%x (%d) is being shared"
+   " between thread %lu and current  %lu",
+   dpaa2_io_portal[lcore_id].sec_dpio_dev,
+   dpaa2_io_portal[lcore_id].sec_dpio_dev->index,
+   dpaa2_io_portal[lcore_id].sec_tid,
+   tid);
+   RTE_PER_LCORE(_dpaa2_io).sec_dpio_dev
+   = dpaa2_io_portal[lcore_id].sec_dpio_dev;
+   rte_atomic16_inc(&dpaa2_io_portal
+[lcore_id].sec_dpio_dev->ref_count);
+   dpaa2_io_portal[lcore_id].sec_tid = tid;
+
+   PMD_DRV_LOG(DEBUG, "Old Portal=0x%x (%d) affined thread - %lu",
+   dpaa2_io_portal[lcore_id].sec_dpio_dev,
+   dpaa2_io_portal[lcore_id].sec_dpio_dev->index,
+   tid);
+   return 0;
+   }
+
+   /* Populate the dpaa2_io_portal structure */
+   dpaa2_io_portal[lcore_id].sec_dpio_dev = dpaa2_get_qbman_swp();
+
+   if (dpaa2_io_portal[lcore_id].sec_dpio_dev) {
+   RTE_PER_LCORE(_dpaa2_io).sec_dpio_dev
+   = dpaa2_io_portal[lcore_id].sec_dpio_dev;
+   dpaa2_io_portal[lcore_id].sec_tid = tid;
+   return 0;
+   } else {
+   return -1;
+   }
+}
+
+int
 dpaa2_create_dpio_device(struct fslmc_vfio_device *vdev,
 struct vfio_device_info *obj_info,
int object_id)
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
index 682f3fa..b1a1b8f 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
@@ -56,5 +56,8 @@ struct dpaa2_io_portal_t {
 /* Affine a DPIO portal to current processing thread */
 int dpaa2_affine_qbman_swp(void);
 
+/* Affine additional DPIO portal to current crypto processing thread */
+int dpaa2_affine_qbman_swp_sec(void);
+
 
 #endif /* _DPAA2_HW_DPIO_H_ */
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index 028f55e..4a8f478 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -2,6 +2,7 @@ DPDK_17.02 {
global:
 
 dpaa2_affine_qbman_swp;
+dpaa2_affine_qbman_swp_sec;
 dpaa2_alloc_dpbp_dev;
 dpaa2_free_dpbp_dev;
 dpbp_disable;
-- 
1.9.1

[dpdk-dev] [PATCHv6 18/33] net/dpaa2: adding eth ops to dpaa2

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini  |   1 +
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h |  20 ++
 drivers/net/dpaa2/Makefile  |   3 +
 drivers/net/dpaa2/dpaa2_ethdev.c| 412 +++-
 drivers/net/dpaa2/dpaa2_ethdev.h|  15 ++
 5 files changed, 450 insertions(+), 1 deletion(-)

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index b176208..0b59725 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -4,6 +4,7 @@
 ; Refer to default.ini for the full list of available PMD features.
 ;
 [Features]
+Queue start/stop = Y
 Linux VFIO   = Y
 ARMv8= Y
 Usage doc= Y
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index ad2847f..42c5517 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -37,9 +37,12 @@
 #include 
 #include 
 
+#define DPAA2_DQRR_RING_SIZE   16
+   /** 
 #include 
+#include 
+
 #include "dpaa2_ethdev.h"
 
 static struct rte_dpaa2_driver rte_dpaa2_pmd;
 
+static void
+dpaa2_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
+{
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+
+   PMD_INIT_FUNC_TRACE();
+
+   dev_info->if_index = priv->hw_id;
+
+   dev_info->max_rx_queues = (uint16_t)priv->nb_rx_queues;
+   dev_info->max_tx_queues = (uint16_t)priv->nb_tx_queues;
+
+   dev_info->speed_capa = ETH_LINK_SPEED_1G |
+   ETH_LINK_SPEED_2_5G |
+   ETH_LINK_SPEED_10G;
+}
+
+static int
+dpaa2_alloc_rx_tx_queues(struct rte_eth_dev *dev)
+{
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+   uint16_t dist_idx;
+   uint32_t vq_id;
+   struct dpaa2_queue *mc_q, *mcq;
+   uint32_t tot_queues;
+   int i;
+   struct dpaa2_queue *dpaa2_q;
+
+   PMD_INIT_FUNC_TRACE();
+
+   tot_queues = priv->nb_rx_queues + priv->nb_tx_queues;
+   mc_q = rte_malloc(NULL, sizeof(struct dpaa2_queue) * tot_queues,
+ RTE_CACHE_LINE_SIZE);
+   if (!mc_q) {
+   PMD_INIT_LOG(ERR, "malloc failed for rx/tx queues\n");
+   return -1;
+   }
+
+   for (i = 0; i < priv->nb_rx_queues; i++) {
+   mc_q->dev = dev;
+   priv->rx_vq[i] = mc_q++;
+   dpaa2_q = (struct dpaa2_queue *)priv->rx_vq[i];
+   dpaa2_q->q_storage = rte_malloc("dq_storage",
+   sizeof(struct queue_storage_info_t),
+   RTE_CACHE_LINE_SIZE);
+   if (!dpaa2_q->q_storage)
+   goto fail;
+
+   memset(dpaa2_q->q_storage, 0,
+  sizeof(struct queue_storage_info_t));
+   dpaa2_q->q_storage->dq_storage[0] = rte_malloc(NULL,
+   DPAA2_DQRR_RING_SIZE * sizeof(struct qbman_result),
+   RTE_CACHE_LINE_SIZE);
+   }
+
+   for (i = 0; i < priv->nb_tx_queues; i++) {
+   mc_q->dev = dev;
+   mc_q->flow_id = DPNI_NEW_FLOW_ID;
+   priv->tx_vq[i] = mc_q++;
+   }
+
+   vq_id = 0;
+   for (dist_idx = 0; dist_idx < priv->nb_rx_queues; dist_idx++) {
+   mcq = (struct dpaa2_queue *)priv->rx_vq[vq_id];
+   mcq->tc_index = DPAA2_DEF_TC;
+   mcq->flow_id = dist_idx;
+   vq_id++;
+   }
+
+   return 0;
+fail:
+   i -= 1;
+   mc_q = priv->rx_vq[0];
+   while (i >= 0) {
+   dpaa2_q = (struct dpaa2_queue *)priv->rx_vq[i];
+   rte_free(dpaa2_q->q_storage->dq_storage[0]);
+   rte_free(dpaa2_q->q_storage);
+   priv->rx_vq[i--] = NULL;
+   }
+   rte_free(mc_q);
+   return -1;
+}
+
+static int
+dpaa2_eth_dev_configure(struct rte_eth_dev *dev)
+{
+   struct rte_eth_dev_data *data = dev->data;
+   struct rte_eth_conf *eth_conf = &data->dev_conf;
+
+   PMD_INIT_FUNC_TRACE();
+
+   /* Check for correct configuration */
+   if (eth_conf->rxmode.mq_mode != ETH_MQ_RX_RSS &&
+   data->nb_rx_queues > 1) {
+   PMD_INIT_LOG(ERR, "Distribution is not enabled, "
+   "but Rx queues more than 1\n");
+   return -1;
+   }
+
+   return 0;
+}
+
+/* Function to setup RX flow information. It contains traffic class ID,
+ * flow ID, destination configuration etc.
+ */
+static int
+dpaa2_dev_rx_queue_setup(struct rte_eth_dev *dev,
+uint16_t rx_queue_id,
+uint16_t nb_rx_desc __rte_unused,
+unsigned int socket_id __rte_unused,
+const struct rte_eth_rxconf *rx_conf __rte_unused,
+struct rte_mempool *mb_pool)
+{
+   struct dpaa2_dev_

[dpdk-dev] [PATCHv6 19/33] net/dpaa2: add rss flow distribution

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini |   1 +
 drivers/net/dpaa2/Makefile |   1 +
 drivers/net/dpaa2/base/dpaa2_hw_dpni.c | 287 +
 drivers/net/dpaa2/dpaa2_ethdev.c   |  31 +++-
 drivers/net/dpaa2/dpaa2_ethdev.h   |  12 ++
 5 files changed, 328 insertions(+), 4 deletions(-)
 create mode 100644 drivers/net/dpaa2/base/dpaa2_hw_dpni.c

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index 0b59725..20152a0 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -5,6 +5,7 @@
 ;
 [Features]
 Queue start/stop = Y
+RSS hash = Y
 Linux VFIO   = Y
 ARMv8= Y
 Usage doc= Y
diff --git a/drivers/net/dpaa2/Makefile b/drivers/net/dpaa2/Makefile
index 5fb5f67..7d62f1e 100644
--- a/drivers/net/dpaa2/Makefile
+++ b/drivers/net/dpaa2/Makefile
@@ -57,6 +57,7 @@ EXPORT_MAP := rte_pmd_dpaa2_version.map
 # library version
 LIBABIVER := 1
 
+SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += base/dpaa2_hw_dpni.c
 SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_ethdev.c
 
 # library dependencies
diff --git a/drivers/net/dpaa2/base/dpaa2_hw_dpni.c 
b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
new file mode 100644
index 000..c95c083
--- /dev/null
+++ b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
@@ -0,0 +1,287 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright (c) 2016 Freescale Semiconductor, Inc. All rights reserved.
+ *   Copyright (c) 2016 NXP. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Freescale Semiconductor, Inc nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+
+#include "../dpaa2_ethdev.h"
+
+static void
+dpaa2_distset_to_dpkg_profile_cfg(
+   uint32_t req_dist_set,
+   struct dpkg_profile_cfg *kg_cfg);
+
+int
+dpaa2_setup_flow_dist(struct rte_eth_dev *eth_dev,
+ uint32_t req_dist_set)
+{
+   struct dpaa2_dev_priv *priv = eth_dev->data->dev_private;
+   struct fsl_mc_io *dpni = priv->hw;
+   struct dpni_rx_tc_dist_cfg tc_cfg;
+   struct dpkg_profile_cfg kg_cfg;
+   void *p_params;
+   int ret, tc_index = 0;
+
+   p_params = rte_malloc(
+   NULL, DIST_PARAM_IOVA_SIZE, RTE_CACHE_LINE_SIZE);
+   if (!p_params) {
+   RTE_LOG(ERR, PMD, "Memory unavaialble\n");
+   return -ENOMEM;
+   }
+   memset(p_params, 0, DIST_PARAM_IOVA_SIZE);
+   memset(&tc_cfg, 0, sizeof(struct dpni_rx_tc_dist_cfg));
+
+   dpaa2_distset_to_dpkg_profile_cfg(req_dist_set, &kg_cfg);
+   tc_cfg.key_cfg_iova = (uint64_t)(p_params);
+   tc_cfg.dist_size = eth_dev->data->nb_rx_queues;
+   tc_cfg.dist_mode = DPNI_DIST_MODE_HASH;
+
+   ret = dpni_prepare_key_cfg(&kg_cfg, p_params);
+   if (ret) {
+   RTE_LOG(ERR, PMD, "Unable to prepare extract parameters\n");
+   rte_free(p_params);
+   return ret;
+   }
+
+   ret = dpni_set_rx_tc_dist(dpni, CMD_PRI_LOW, priv->token, tc_index,
+ &tc_cfg);
+   rte_free(p_params);
+   if (ret) {
+   RTE_LOG(ERR, PMD, "Setting distribution for Rx failed with"
+   " err code: %d\n", ret);
+   return ret;
+   }
+
+   return 0;
+}
+
+int dpaa2

[dpdk-dev] [PATCHv6 21/33] net/dpaa2: attach the buffer pool to dpni

[Hemant Agrawal]

This patch configures a MC-DPNI based DPAA2 PMD network
port with a DPBP based buffer pool.

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 10 ++
 drivers/net/dpaa2/Makefile  |  3 ++
 drivers/net/dpaa2/base/dpaa2_hw_dpni.c  | 57 ++
 drivers/net/dpaa2/dpaa2_ethdev.c| 62 +
 drivers/net/dpaa2/dpaa2_ethdev.h|  6 
 5 files changed, 138 insertions(+)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 42c5517..8efac2d 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -50,6 +50,16 @@
 #define DPAA2_MBUF_MAX_ACQ_REL 7
 
 #define MAX_BPID 256
+#define DPAA2_MBUF_HW_ANNOTATION   64
+#define DPAA2_FD_PTA_SIZE  64
+
+#if (DPAA2_MBUF_HW_ANNOTATION + DPAA2_FD_PTA_SIZE) > RTE_PKTMBUF_HEADROOM
+#error "Annotation requirement is more than RTE_PKTMBUF_HEADROOM"
+#endif
+
+/* we will re-use the HEADROOM for annotation in RX */
+#define DPAA2_HW_BUF_RESERVE   0
+#define DPAA2_PACKET_LAYOUT_ALIGN  64 /*changing from 256 */
 
 struct dpaa2_dpio_dev {
TAILQ_ENTRY(dpaa2_dpio_dev) next;
diff --git a/drivers/net/dpaa2/Makefile b/drivers/net/dpaa2/Makefile
index 7d62f1e..fadabb5 100644
--- a/drivers/net/dpaa2/Makefile
+++ b/drivers/net/dpaa2/Makefile
@@ -49,6 +49,7 @@ CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/mc
 CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/portal
 CFLAGS += -I$(RTE_SDK)/drivers/common/dpaa2/qbman/include
+CFLAGS += -I$(RTE_SDK)/drivers/pool/dpaa2
 CFLAGS += -I$(RTE_SDK)/lib/librte_eal/linuxapp/eal
 
 # versioning export map
@@ -62,7 +63,9 @@ SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_ethdev.c
 
 # library dependencies
 DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_eal lib/librte_ether
+DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_mempool lib/librte_mbuf
 DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_common_dpaa2_qbman
 DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_bus_fslmc
+DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_pmd_dpaa2_pool
 
 include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/net/dpaa2/base/dpaa2_hw_dpni.c 
b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
index c95c083..08f53b3 100644
--- a/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
+++ b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
@@ -46,6 +46,7 @@
 
 #include 
 #include 
+#include 
 
 #include "../dpaa2_ethdev.h"
 
@@ -285,3 +286,59 @@ int dpaa2_remove_flow_dist(
}
kg_cfg->num_extracts = i;
 }
+
+int
+dpaa2_attach_bp_list(struct dpaa2_dev_priv *priv,
+void *blist)
+{
+   /* Function to attach a DPNI with a buffer pool list. Buffer pool list
+* handle is passed in blist.
+*/
+   int32_t retcode;
+   struct fsl_mc_io *dpni = priv->hw;
+   struct dpni_pools_cfg bpool_cfg;
+   struct dpaa2_bp_list *bp_list = (struct dpaa2_bp_list *)blist;
+   struct dpni_buffer_layout layout;
+   int tot_size;
+
+   /* ... rx buffer layout .
+* Check alignment for buffer layouts first
+*/
+
+   /* ... rx buffer layout ... */
+   tot_size = DPAA2_HW_BUF_RESERVE + RTE_PKTMBUF_HEADROOM;
+   tot_size = RTE_ALIGN_CEIL(tot_size,
+ DPAA2_PACKET_LAYOUT_ALIGN);
+
+   memset(&layout, 0, sizeof(struct dpni_buffer_layout));
+   layout.options = DPNI_BUF_LAYOUT_OPT_DATA_HEAD_ROOM;
+
+   layout.data_head_room =
+   tot_size - DPAA2_FD_PTA_SIZE - DPAA2_MBUF_HW_ANNOTATION;
+   retcode = dpni_set_buffer_layout(dpni, CMD_PRI_LOW, priv->token,
+DPNI_QUEUE_RX, &layout);
+   if (retcode) {
+   PMD_INIT_LOG(ERR, "Err(%d) in setting rx buffer layout\n",
+retcode);
+   return retcode;
+   }
+
+   /*Attach buffer pool to the network interface as described by the user*/
+   bpool_cfg.num_dpbp = 1;
+   bpool_cfg.pools[0].dpbp_id = bp_list->buf_pool.dpbp_node->dpbp_id;
+   bpool_cfg.pools[0].backup_pool = 0;
+   bpool_cfg.pools[0].buffer_size =
+   RTE_ALIGN_CEIL(bp_list->buf_pool.size,
+  256 /*DPAA2_PACKET_LAYOUT_ALIGN*/);
+
+   retcode = dpni_set_pools(dpni, CMD_PRI_LOW, priv->token, &bpool_cfg);
+   if (retcode != 0) {
+   PMD_INIT_LOG(ERR, "Error in attaching the buffer pool list"
+   " bpid = %d Error code = %d\n",
+   bpool_cfg.pools[0].dpbp_id, retcode);
+   return retcode;
+   }
+
+   priv->bp_list = bp_list;
+   return 0;
+}
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 979dba4..6de571a 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -48,6 +48,7 @@
 #include 
 #include 
 #i

[dpdk-dev] [PATCHv6 20/33] net/dpaa2: configure mac address at init

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 drivers/net/dpaa2/dpaa2_ethdev.c | 28 
 drivers/net/dpaa2/dpaa2_ethdev.h |  3 +++
 2 files changed, 31 insertions(+)

diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 87f7640..979dba4 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -62,6 +62,7 @@
 
dev_info->if_index = priv->hw_id;
 
+   dev_info->max_mac_addrs = priv->max_mac_filters;
dev_info->max_rx_queues = (uint16_t)priv->nb_rx_queues;
dev_info->max_tx_queues = (uint16_t)priv->nb_tx_queues;
 
@@ -446,6 +447,9 @@
 
priv->hw = dpni_dev;
priv->hw_id = hw_id;
+   priv->options = attr.options;
+   priv->max_mac_filters = attr.mac_filter_entries;
+   priv->max_vlan_filters = attr.vlan_filter_entries;
priv->flags = 0;
 
ret = dpaa2_alloc_rx_tx_queues(eth_dev);
@@ -454,6 +458,25 @@
return -ret;
}
 
+   /* Allocate memory for storing MAC addresses */
+   eth_dev->data->mac_addrs = rte_zmalloc("dpni",
+   ETHER_ADDR_LEN * attr.mac_filter_entries, 0);
+   if (eth_dev->data->mac_addrs == NULL) {
+   PMD_INIT_LOG(ERR, "Failed to allocate %d bytes needed to "
+   "store MAC addresses",
+   ETHER_ADDR_LEN * attr.mac_filter_entries);
+   return -ENOMEM;
+   }
+
+   ret = dpni_get_primary_mac_addr(dpni_dev, CMD_PRI_LOW,
+   priv->token,
+   (uint8_t *)(eth_dev->data->mac_addrs[0].addr_bytes));
+   if (ret) {
+   PMD_INIT_LOG(ERR, "DPNI get mac address failed:"
+   " Error Code = %d\n", ret);
+   return -ret;
+   }
+
eth_dev->dev_ops = &dpaa2_ethdev_ops;
eth_dev->data->drv_name = rte_dpaa2_pmd.driver.name;
 
@@ -492,6 +515,11 @@
priv->rx_vq[0] = NULL;
}
 
+   /* Allocate memory for storing MAC addresses */
+   if (eth_dev->data->mac_addrs) {
+   rte_free(eth_dev->data->mac_addrs);
+   eth_dev->data->mac_addrs = NULL;
+   }
 
/*Close the device at underlying layer*/
ret = dpni_close(dpni, CMD_PRI_LOW, priv->token);
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.h b/drivers/net/dpaa2/dpaa2_ethdev.h
index d24fcc6..2d13137 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.h
+++ b/drivers/net/dpaa2/dpaa2_ethdev.h
@@ -57,7 +57,10 @@ struct dpaa2_dev_priv {
void *rx_vq[MAX_RX_QUEUES];
void *tx_vq[MAX_TX_QUEUES];
 
+   uint32_t options;
uint16_t num_dist_per_tc[MAX_TCS];
+   uint8_t max_mac_filters;
+   uint8_t max_vlan_filters;
uint8_t num_tc;
uint8_t flags; /*dpaa2 config flags */
 };
-- 
1.9.1

[dpdk-dev] [PATCHv6 22/33] net/dpaa2: add support for l3 and l4 checksum offload

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini  |  2 +
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h |  6 +++
 drivers/net/dpaa2/Makefile  |  2 +-
 drivers/net/dpaa2/dpaa2_ethdev.c| 72 +++--
 4 files changed, 77 insertions(+), 5 deletions(-)

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index 20152a0..d50c62e 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -6,6 +6,8 @@
 [Features]
 Queue start/stop = Y
 RSS hash = Y
+L3 checksum offload  = Y
+L4 checksum offload  = Y
 Linux VFIO   = Y
 ARMv8= Y
 Usage doc= Y
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 8efac2d..1af93a5 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -37,6 +37,12 @@
 #include 
 #include 
 
+#ifndef false
+#define false  0
+#endif
+#ifndef true
+#define true   1
+#endif
 #define DPAA2_DQRR_RING_SIZE   16
/** min_rx_bufsize = DPAA2_MIN_RX_BUF_SIZE;
dev_info->max_rx_queues = (uint16_t)priv->nb_rx_queues;
dev_info->max_tx_queues = (uint16_t)priv->nb_tx_queues;
-
+   dev_info->rx_offload_capa =
+   DEV_RX_OFFLOAD_IPV4_CKSUM |
+   DEV_RX_OFFLOAD_UDP_CKSUM |
+   DEV_RX_OFFLOAD_TCP_CKSUM |
+   DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM;
+   dev_info->tx_offload_capa =
+   DEV_TX_OFFLOAD_IPV4_CKSUM |
+   DEV_TX_OFFLOAD_UDP_CKSUM |
+   DEV_TX_OFFLOAD_TCP_CKSUM |
+   DEV_TX_OFFLOAD_SCTP_CKSUM |
+   DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM;
dev_info->speed_capa = ETH_LINK_SPEED_1G |
ETH_LINK_SPEED_2_5G |
ETH_LINK_SPEED_10G;
@@ -252,8 +262,13 @@
memset(&tx_conf_cfg, 0, sizeof(struct dpni_queue));
memset(&tx_flow_cfg, 0, sizeof(struct dpni_queue));
 
-   tc_id = 0;
-   flow_id = tx_queue_id;
+   if (priv->num_tc == 1) {
+   tc_id = 0;
+   flow_id = tx_queue_id % priv->num_dist_per_tc[tc_id];
+   } else {
+   tc_id = tx_queue_id;
+   flow_id = 0;
+   }
 
ret = dpni_set_queue(dpni, CMD_PRI_LOW, priv->token, DPNI_QUEUE_TX,
 tc_id, flow_id, options, &tx_flow_cfg);
@@ -302,6 +317,7 @@
struct dpaa2_dev_priv *priv = data->dev_private;
struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
struct dpni_queue cfg;
+   struct dpni_error_cfg   err_cfg;
uint16_t qdid;
struct dpni_queue_id qid;
struct dpaa2_queue *dpaa2_q;
@@ -337,6 +353,48 @@
dpaa2_q->fqid = qid.fqid;
}
 
+   ret = dpni_set_offload(dpni, CMD_PRI_LOW, priv->token,
+  DPNI_OFF_RX_L3_CSUM, true);
+   if (ret) {
+   PMD_INIT_LOG(ERR, "Error to set RX l3 csum:Error = %d\n", ret);
+   return ret;
+   }
+
+   ret = dpni_set_offload(dpni, CMD_PRI_LOW, priv->token,
+  DPNI_OFF_RX_L4_CSUM, true);
+   if (ret) {
+   PMD_INIT_LOG(ERR, "Error to get RX l4 csum:Error = %d\n", ret);
+   return ret;
+   }
+
+   ret = dpni_set_offload(dpni, CMD_PRI_LOW, priv->token,
+  DPNI_OFF_TX_L3_CSUM, true);
+   if (ret) {
+   PMD_INIT_LOG(ERR, "Error to set TX l3 csum:Error = %d\n", ret);
+   return ret;
+   }
+
+   ret = dpni_set_offload(dpni, CMD_PRI_LOW, priv->token,
+  DPNI_OFF_TX_L4_CSUM, true);
+   if (ret) {
+   PMD_INIT_LOG(ERR, "Error to get TX l4 csum:Error = %d\n", ret);
+   return ret;
+   }
+
+   /*checksum errors, send them to normal path and set it in annotation */
+   err_cfg.errors = DPNI_ERROR_L3CE | DPNI_ERROR_L4CE;
+
+   err_cfg.error_action = DPNI_ERROR_ACTION_CONTINUE;
+   err_cfg.set_frame_annotation = true;
+
+   ret = dpni_set_errors_behavior(dpni, CMD_PRI_LOW,
+  priv->token, &err_cfg);
+   if (ret) {
+   PMD_INIT_LOG(ERR, "Error to dpni_set_errors_behavior:"
+"code = %d\n", ret);
+   return ret;
+   }
+
return 0;
 }
 
@@ -453,7 +511,13 @@
 */
priv->nb_rx_queues = priv->num_dist_per_tc[DPAA2_DEF_TC];
 
-   priv->nb_tx_queues = attr.num_queues;
+   if (attr.num_tcs == 1)
+   priv->nb_tx_queues = attr.num_queues;
+   else
+   priv->nb_tx_queues = attr.num_tcs;
+
+   PMD_INIT_LOG(DEBUG, "num_tc %d", priv->num_tc);
+   PMD_INIT_LOG(DEBUG, "nb_rx_queues %d", priv->nb_rx_queues);
 
eth_dev->data->nb_rx_queues = priv->nb_rx_queues;
eth_dev->data->nb_tx_queues = priv->nb_tx_que

[dpdk-dev] [PATCHv6 23/33] net/dpaa2: add support for promiscuous mode

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini |  1 +
 drivers/net/dpaa2/dpaa2_ethdev.c   | 41 ++
 2 files changed, 42 insertions(+)

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index d50c62e..b7c274a 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -5,6 +5,7 @@
 ;
 [Features]
 Queue start/stop = Y
+Promiscuous mode = Y
 RSS hash = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 2298246..df78f6c 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -437,11 +437,52 @@
}
 }
 
+static void
+dpaa2_dev_promiscuous_enable(
+   struct rte_eth_dev *dev)
+{
+   int ret;
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+   struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
+
+   PMD_INIT_FUNC_TRACE();
+
+   if (dpni == NULL) {
+   RTE_LOG(ERR, PMD, "dpni is NULL");
+   return;
+   }
+
+   ret = dpni_set_unicast_promisc(dpni, CMD_PRI_LOW, priv->token, true);
+   if (ret < 0)
+   RTE_LOG(ERR, PMD, "Unable to enable promiscuous mode %d", ret);
+}
+
+static void
+dpaa2_dev_promiscuous_disable(
+   struct rte_eth_dev *dev)
+{
+   int ret;
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+   struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
+
+   PMD_INIT_FUNC_TRACE();
+
+   if (dpni == NULL) {
+   RTE_LOG(ERR, PMD, "dpni is NULL");
+   return;
+   }
+
+   ret = dpni_set_unicast_promisc(dpni, CMD_PRI_LOW, priv->token, false);
+   if (ret < 0)
+   RTE_LOG(ERR, PMD, "Unable to disable promiscuous mode %d", ret);
+}
 static struct eth_dev_ops dpaa2_ethdev_ops = {
.dev_configure= dpaa2_eth_dev_configure,
.dev_start= dpaa2_dev_start,
.dev_stop = dpaa2_dev_stop,
.dev_close= dpaa2_dev_close,
+   .promiscuous_enable   = dpaa2_dev_promiscuous_enable,
+   .promiscuous_disable  = dpaa2_dev_promiscuous_disable,
.dev_infos_get = dpaa2_dev_info_get,
.rx_queue_setup= dpaa2_dev_rx_queue_setup,
.rx_queue_release  = dpaa2_dev_rx_queue_release,
-- 
1.9.1

[dpdk-dev] [PATCHv6 25/33] net/dpaa2: add packet rx and tx support

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h |  53 +++
 drivers/net/dpaa2/Makefile  |   1 +
 drivers/net/dpaa2/dpaa2_ethdev.c|   4 +
 drivers/net/dpaa2/dpaa2_ethdev.h|   3 +
 drivers/net/dpaa2/dpaa2_rxtx.c  | 260 
 5 files changed, 321 insertions(+)
 create mode 100644 drivers/net/dpaa2/dpaa2_rxtx.c

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 2a8d9e5..c26360d3 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -43,10 +43,16 @@
 #ifndef true
 #define true   1
 #endif
+#define lower_32_bits(x) ((uint32_t)(x))
+#define upper_32_bits(x) ((uint32_t)(((x) >> 16) >> 16))
 
 #ifndef ETH_VLAN_HLEN
 #define ETH_VLAN_HLEN   4 /** < Vlan Header Length */
 #endif
+
+#define MAX_TX_RING_SLOTS  8
+   /** simple.addr_lo = lower_32_bits((uint64_t)(addr));   \
+   fd->simple.addr_hi = upper_32_bits((uint64_t)(addr));   \
+} while (0)
+#define DPAA2_SET_FD_LEN(fd, length)   (fd)->simple.len = length
+#define DPAA2_SET_FD_BPID(fd, bpid)((fd)->simple.bpid_offset |= bpid)
+#define DPAA2_SET_FD_OFFSET(fd, offset)\
+   ((fd->simple.bpid_offset |= (uint32_t)(offset) << 16))
+#define DPAA2_RESET_FD_CTRL(fd)(fd)->simple.ctrl = 0
+
+#defineDPAA2_SET_FD_ASAL(fd, asal) ((fd)->simple.ctrl |= (asal << 
16))
+#define DPAA2_SET_FD_FLC(fd, addr) do { \
+   fd->simple.flc_lo = lower_32_bits((uint64_t)(addr));\
+   fd->simple.flc_hi = upper_32_bits((uint64_t)(addr));\
+} while (0)
+#define DPAA2_GET_FD_ADDR(fd)  \
+((uint64_t)uint64_t)((fd)->simple.addr_hi)) << 32) + (fd)->simple.addr_lo))
+
+#define DPAA2_GET_FD_LEN(fd)   ((fd)->simple.len)
+#define DPAA2_GET_FD_BPID(fd)  (((fd)->simple.bpid_offset & 0x3FFF))
+#define DPAA2_GET_FD_OFFSET(fd)(((fd)->simple.bpid_offset & 
0x0FFF) >> 16)
+#define DPAA2_INLINE_MBUF_FROM_BUF(buf, meta_data_size) \
+   ((struct rte_mbuf *)((uint64_t)(buf) - (meta_data_size)))
+
+#define DPAA2_ASAL_VAL (DPAA2_MBUF_HW_ANNOTATION / 64)
+
+/* Only Enqueue Error responses will be
+ * pushed on FQID_ERR of Enqueue FQ
+ */
+#define DPAA2_EQ_RESP_ERR_FQ   0
+/* All Enqueue responses will be pushed on address
+ * set with qbman_eq_desc_set_response
+ */
+#define DPAA2_EQ_RESP_ALWAYS   1
+
 struct dpaa2_dpbp_dev *dpaa2_alloc_dpbp_dev(void);
 void dpaa2_free_dpbp_dev(struct dpaa2_dpbp_dev *dpbp);
 
diff --git a/drivers/net/dpaa2/Makefile b/drivers/net/dpaa2/Makefile
index 0e59203..d52fa39 100644
--- a/drivers/net/dpaa2/Makefile
+++ b/drivers/net/dpaa2/Makefile
@@ -59,6 +59,7 @@ EXPORT_MAP := rte_pmd_dpaa2_version.map
 LIBABIVER := 1
 
 SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += base/dpaa2_hw_dpni.c
+SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_rxtx.c
 SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_ethdev.c
 
 # library dependencies
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 7e64e60..c2015f0 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -681,6 +681,8 @@
eth_dev->dev_ops = &dpaa2_ethdev_ops;
eth_dev->data->drv_name = rte_dpaa2_pmd.driver.name;
 
+   eth_dev->rx_pkt_burst = dpaa2_dev_rx;
+   eth_dev->tx_pkt_burst = dpaa2_dev_tx;
return 0;
 }
 
@@ -734,6 +736,8 @@
free(dpni);
 
eth_dev->dev_ops = NULL;
+   eth_dev->rx_pkt_burst = NULL;
+   eth_dev->tx_pkt_burst = NULL;
 
return 0;
 }
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.h b/drivers/net/dpaa2/dpaa2_ethdev.h
index a56b525..7196398 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.h
+++ b/drivers/net/dpaa2/dpaa2_ethdev.h
@@ -77,4 +77,7 @@ int dpaa2_remove_flow_dist(struct rte_eth_dev *eth_dev,
 
 int dpaa2_attach_bp_list(struct dpaa2_dev_priv *priv, void *blist);
 
+uint16_t dpaa2_dev_rx(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts);
+uint16_t dpaa2_dev_tx(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts);
+
 #endif /* _DPAA2_ETHDEV_H */
diff --git a/drivers/net/dpaa2/dpaa2_rxtx.c b/drivers/net/dpaa2/dpaa2_rxtx.c
new file mode 100644
index 000..25574c0
--- /dev/null
+++ b/drivers/net/dpaa2/dpaa2_rxtx.c
@@ -0,0 +1,260 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright (c) 2016 Freescale Semiconductor, Inc. All rights reserved.
+ *   Copyright (c) 2016 NXP. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *  

[dpdk-dev] [PATCHv6 26/33] net/dpaa2: rx packet parsing and packet type support

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini   |   1 +
 drivers/net/dpaa2/base/dpaa2_hw_dpni_annot.h | 257 +++
 drivers/net/dpaa2/dpaa2_ethdev.c |  23 +++
 drivers/net/dpaa2/dpaa2_rxtx.c   |  91 +-
 4 files changed, 371 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/dpaa2/base/dpaa2_hw_dpni_annot.h

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index a6b7964..0746d4b 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -10,6 +10,7 @@ Promiscuous mode = Y
 RSS hash = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
+Packet type parsing  = Y
 Linux VFIO   = Y
 ARMv8= Y
 Usage doc= Y
diff --git a/drivers/net/dpaa2/base/dpaa2_hw_dpni_annot.h 
b/drivers/net/dpaa2/base/dpaa2_hw_dpni_annot.h
new file mode 100644
index 000..9324c6a
--- /dev/null
+++ b/drivers/net/dpaa2/base/dpaa2_hw_dpni_annot.h
@@ -0,0 +1,257 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright (c) 2016 Freescale Semiconductor, Inc. All rights reserved.
+ *   Copyright (c) 2016 NXP. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Freescale Semiconductor, Inc nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/**
+ * @file
+ *
+ * DPNI packet parse results - implementation internal
+ */
+
+#ifndef _DPAA2_HW_DPNI_ANNOT_H_
+#define _DPAA2_HW_DPNI_ANNOT_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* Annotation valid bits in FD FRC */
+#define DPAA2_FD_FRC_FASV  0x8000
+#define DPAA2_FD_FRC_FAEADV0x4000
+#define DPAA2_FD_FRC_FAPRV 0x2000
+#define DPAA2_FD_FRC_FAIADV0x1000
+#define DPAA2_FD_FRC_FASWOV0x0800
+#define DPAA2_FD_FRC_FAICFDV   0x0400
+
+/* Annotation bits in FD CTRL */
+#define DPAA2_FD_CTRL_ASAL 0x0002  /* ASAL = 128 */
+#define DPAA2_FD_CTRL_PTA  0x0080
+#define DPAA2_FD_CTRL_PTV1 0x0040
+
+/* Frame annotation status */
+struct dpaa2_fas {
+   uint8_t reserved;
+   uint8_t ppid;
+   __le16 ifpid;
+   __le32 status;
+} __packed;
+
+/**
+ * HW Packet Annotation  Register structures
+ */
+struct dpaa2_annot_hdr {
+   /**

[dpdk-dev] [PATCHv6 24/33] net/dpaa2: add mtu config support

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini  |  1 +
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h |  4 
 drivers/net/dpaa2/dpaa2_ethdev.c| 34 +
 3 files changed, 39 insertions(+)

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index b7c274a..a6b7964 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -5,6 +5,7 @@
 ;
 [Features]
 Queue start/stop = Y
+MTU update   = Y
 Promiscuous mode = Y
 RSS hash = Y
 L3 checksum offload  = Y
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 1af93a5..2a8d9e5 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -43,6 +43,10 @@
 #ifndef true
 #define true   1
 #endif
+
+#ifndef ETH_VLAN_HLEN
+#define ETH_VLAN_HLEN   4 /** < Vlan Header Length */
+#endif
 #define DPAA2_DQRR_RING_SIZE   16
/** data->dev_private;
+   struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
+   uint32_t frame_size = mtu + ETHER_HDR_LEN + ETHER_CRC_LEN;
+
+   PMD_INIT_FUNC_TRACE();
+
+   if (dpni == NULL) {
+   RTE_LOG(ERR, PMD, "dpni is NULL");
+   return -EINVAL;
+   }
+
+   /* check that mtu is within the allowed range */
+   if ((mtu < ETHER_MIN_MTU) || (frame_size > DPAA2_MAX_RX_PKT_LEN))
+   return -EINVAL;
+
+   /* Set the Max Rx frame length as 'mtu' +
+* Maximum Ethernet header length
+*/
+   ret = dpni_set_max_frame_length(dpni, CMD_PRI_LOW, priv->token,
+   mtu + ETH_VLAN_HLEN);
+   if (ret) {
+   PMD_DRV_LOG(ERR, "setting the max frame length failed");
+   return -1;
+   }
+   PMD_DRV_LOG(INFO, "MTU is configured %d for the device\n", mtu);
+   return 0;
+}
+
 static struct eth_dev_ops dpaa2_ethdev_ops = {
.dev_configure= dpaa2_eth_dev_configure,
.dev_start= dpaa2_dev_start,
@@ -484,6 +517,7 @@
.promiscuous_enable   = dpaa2_dev_promiscuous_enable,
.promiscuous_disable  = dpaa2_dev_promiscuous_disable,
.dev_infos_get = dpaa2_dev_info_get,
+   .mtu_set   = dpaa2_dev_mtu_set,
.rx_queue_setup= dpaa2_dev_rx_queue_setup,
.rx_queue_release  = dpaa2_dev_rx_queue_release,
.tx_queue_setup= dpaa2_dev_tx_queue_setup,
-- 
1.9.1

[dpdk-dev] [PATCHv6 27/33] net/dpaa2: link status update

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini |   1 +
 drivers/net/dpaa2/dpaa2_ethdev.c   | 107 +
 2 files changed, 108 insertions(+)

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index 0746d4b..0660cab 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -4,6 +4,7 @@
 ; Refer to default.ini for the full list of available PMD features.
 ;
 [Features]
+Link status  = Y
 Queue start/stop = Y
 MTU update   = Y
 Promiscuous mode = Y
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 521a066..4e08096 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -54,6 +54,58 @@
 
 static struct rte_dpaa2_driver rte_dpaa2_pmd;
 
+/**
+ * Atomically reads the link status information from global
+ * structure rte_eth_dev.
+ *
+ * @param dev
+ *   - Pointer to the structure rte_eth_dev to read from.
+ *   - Pointer to the buffer to be saved with the link status.
+ *
+ * @return
+ *   - On success, zero.
+ *   - On failure, negative value.
+ */
+static inline int
+dpaa2_dev_atomic_read_link_status(struct rte_eth_dev *dev,
+ struct rte_eth_link *link)
+{
+   struct rte_eth_link *dst = link;
+   struct rte_eth_link *src = &dev->data->dev_link;
+
+   if (rte_atomic64_cmpset((uint64_t *)dst, *(uint64_t *)dst,
+   *(uint64_t *)src) == 0)
+   return -1;
+
+   return 0;
+}
+
+/**
+ * Atomically writes the link status information into global
+ * structure rte_eth_dev.
+ *
+ * @param dev
+ *   - Pointer to the structure rte_eth_dev to read from.
+ *   - Pointer to the buffer to be saved with the link status.
+ *
+ * @return
+ *   - On success, zero.
+ *   - On failure, negative value.
+ */
+static inline int
+dpaa2_dev_atomic_write_link_status(struct rte_eth_dev *dev,
+  struct rte_eth_link *link)
+{
+   struct rte_eth_link *dst = &dev->data->dev_link;
+   struct rte_eth_link *src = link;
+
+   if (rte_atomic64_cmpset((uint64_t *)dst, *(uint64_t *)dst,
+   *(uint64_t *)src) == 0)
+   return -1;
+
+   return 0;
+}
+
 static void
 dpaa2_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
 {
@@ -430,6 +482,7 @@
struct dpaa2_dev_priv *priv = dev->data->dev_private;
struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
int ret;
+   struct rte_eth_link link;
 
PMD_INIT_FUNC_TRACE();
 
@@ -439,6 +492,10 @@
 ret, priv->hw_id);
return;
}
+
+   /* clear the recorded link status */
+   memset(&link, 0, sizeof(link));
+   dpaa2_dev_atomic_write_link_status(dev, &link);
 }
 
 static void
@@ -531,6 +588,55 @@
return 0;
 }
 
+/* return 0 means link status changed, -1 means not changed */
+static int
+dpaa2_dev_link_update(struct rte_eth_dev *dev,
+   int wait_to_complete __rte_unused)
+{
+   int ret;
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+   struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
+   struct rte_eth_link link, old;
+   struct dpni_link_state state = {0};
+
+   PMD_INIT_FUNC_TRACE();
+
+   if (dpni == NULL) {
+   RTE_LOG(ERR, PMD, "error : dpni is NULL");
+   return 0;
+   }
+   memset(&old, 0, sizeof(old));
+   dpaa2_dev_atomic_read_link_status(dev, &old);
+
+   ret = dpni_get_link_state(dpni, CMD_PRI_LOW, priv->token, &state);
+   if (ret < 0) {
+   RTE_LOG(ERR, PMD, "error: dpni_get_link_state %d", ret);
+   return -1;
+   }
+
+   if ((old.link_status == state.up) && (old.link_speed == state.rate)) {
+   RTE_LOG(DEBUG, PMD, "No change in status\n");
+   return -1;
+   }
+
+   memset(&link, 0, sizeof(struct rte_eth_link));
+   link.link_status = state.up;
+   link.link_speed = state.rate;
+
+   if (state.options & DPNI_LINK_OPT_HALF_DUPLEX)
+   link.link_duplex = ETH_LINK_HALF_DUPLEX;
+   else
+   link.link_duplex = ETH_LINK_FULL_DUPLEX;
+
+   dpaa2_dev_atomic_write_link_status(dev, &link);
+
+   if (link.link_status)
+   PMD_DRV_LOG(INFO, "Port %d Link is Up\n", dev->data->port_id);
+   else
+   PMD_DRV_LOG(INFO, "Port %d Link is Down\n", dev->data->port_id);
+   return 0;
+}
+
 static struct eth_dev_ops dpaa2_ethdev_ops = {
.dev_configure= dpaa2_eth_dev_configure,
.dev_start= dpaa2_dev_start,
@@ -538,6 +644,7 @@
.dev_close= dpaa2_dev_close,
.promiscuous_enable   = dpaa2_dev_promiscuous_enable,
.promiscuous_disable  = dpaa2_dev_promiscuous_disable,
+   .link_update   = dpaa2_

[dpdk-dev] [PATCHv6 30/33] net/dpaa2: add support for non hw buffer pool packet transmit

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 drivers/net/dpaa2/dpaa2_rxtx.c | 75 --
 1 file changed, 73 insertions(+), 2 deletions(-)

diff --git a/drivers/net/dpaa2/dpaa2_rxtx.c b/drivers/net/dpaa2/dpaa2_rxtx.c
index c1ea33a..a94761c 100644
--- a/drivers/net/dpaa2/dpaa2_rxtx.c
+++ b/drivers/net/dpaa2/dpaa2_rxtx.c
@@ -191,6 +191,55 @@ static void __attribute__ ((noinline)) __attribute__((hot))
DPAA2_GET_FD_BPID(fd), DPAA2_GET_FD_LEN(fd));
 }
 
+
+static inline int __attribute__((hot))
+eth_copy_mbuf_to_fd(struct rte_mbuf *mbuf,
+   struct qbman_fd *fd, uint16_t bpid)
+{
+   struct rte_mbuf *m;
+   void *mb = NULL;
+
+   if (rte_dpaa2_mbuf_alloc_bulk(
+   rte_dpaa2_bpid_info[bpid].bp_list->buf_pool.mp, &mb, 1)) {
+   PMD_TX_LOG(WARNING, "Unable to allocated DPAA2 buffer");
+   rte_pktmbuf_free(mbuf);
+   return -1;
+   }
+   m = (struct rte_mbuf *)mb;
+   memcpy((char *)m->buf_addr + mbuf->data_off,
+  (void *)((char *)mbuf->buf_addr + mbuf->data_off),
+   mbuf->pkt_len);
+
+   /* Copy required fields */
+   m->data_off = mbuf->data_off;
+   m->ol_flags = mbuf->ol_flags;
+   m->packet_type = mbuf->packet_type;
+   m->tx_offload = mbuf->tx_offload;
+
+   /*Resetting the buffer pool id and offset field*/
+   fd->simple.bpid_offset = 0;
+
+   DPAA2_SET_FD_ADDR(fd, (m->buf_addr));
+   DPAA2_SET_FD_LEN(fd, mbuf->data_len);
+   DPAA2_SET_FD_BPID(fd, bpid);
+   DPAA2_SET_FD_OFFSET(fd, mbuf->data_off);
+   DPAA2_SET_FD_ASAL(fd, DPAA2_ASAL_VAL);
+
+   PMD_TX_LOG(DEBUG, " mbuf %p BMAN buf addr %p",
+  (void *)mbuf, mbuf->buf_addr);
+
+   PMD_TX_LOG(DEBUG, " fdaddr =%lx bpid =%d meta =%d off =%d, len =%d",
+  DPAA2_GET_FD_ADDR(fd),
+   DPAA2_GET_FD_BPID(fd),
+   rte_dpaa2_bpid_info[DPAA2_GET_FD_BPID(fd)].meta_data_size,
+   DPAA2_GET_FD_OFFSET(fd),
+   DPAA2_GET_FD_LEN(fd));
+   /*free the original packet */
+   rte_pktmbuf_free(mbuf);
+
+   return 0;
+}
+
 uint16_t
 dpaa2_dev_rx(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts)
 {
@@ -331,8 +380,29 @@ static void __attribute__ ((noinline)) __attribute__((hot))
DPAA2_RESET_FD_CTRL((&fd_arr[loop]));
DPAA2_SET_FD_FLC((&fd_arr[loop]), NULL);
mp = (*bufs)->pool;
-   bpid = mempool_to_bpid(mp);
-   eth_mbuf_to_fd(*bufs, &fd_arr[loop], bpid);
+   /* Not a hw_pkt pool allocated frame */
+   if (mp && !(mp->flags & MEMPOOL_F_HW_PKT_POOL)) {
+   PMD_TX_LOG(ERR, "non hw offload bufffer ");
+   /* alloc should be from the default buffer pool
+* attached to this interface
+*/
+   if (priv->bp_list) {
+   bpid = priv->bp_list->buf_pool.bpid;
+   } else {
+   PMD_TX_LOG(ERR, "errr: why no bpool"
+  " attached");
+   num_tx = 0;
+   goto skip_tx;
+   }
+   if (eth_copy_mbuf_to_fd(*bufs,
+   &fd_arr[loop], bpid)) {
+   bufs++;
+   continue;
+   }
+   } else {
+   bpid = mempool_to_bpid(mp);
+   eth_mbuf_to_fd(*bufs, &fd_arr[loop], bpid);
+   }
bufs++;
}
loop = 0;
@@ -345,5 +415,6 @@ static void __attribute__ ((noinline)) __attribute__((hot))
dpaa2_q->tx_pkts += frames_to_send;
nb_pkts -= frames_to_send;
}
+skip_tx:
return num_tx;
 }
-- 
1.9.1

[dpdk-dev] [PATCHv6 28/33] net/dpaa2: basic stats support

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 doc/guides/nics/features/dpaa2.ini |  1 +
 drivers/net/dpaa2/dpaa2_ethdev.c   | 86 ++
 2 files changed, 87 insertions(+)

diff --git a/doc/guides/nics/features/dpaa2.ini 
b/doc/guides/nics/features/dpaa2.ini
index 0660cab..d43f404 100644
--- a/doc/guides/nics/features/dpaa2.ini
+++ b/doc/guides/nics/features/dpaa2.ini
@@ -12,6 +12,7 @@ RSS hash = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 Packet type parsing  = Y
+Basic stats  = Y
 Linux VFIO   = Y
 ARMv8= Y
 Usage doc= Y
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 4e08096..eed0136 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -588,6 +588,90 @@
return 0;
 }
 
+static
+void dpaa2_dev_stats_get(struct rte_eth_dev *dev,
+struct rte_eth_stats *stats)
+{
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+   struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
+   int32_t  retcode;
+   uint8_t page0 = 0, page1 = 1, page2 = 2;
+   union dpni_statistics value;
+
+   memset(&value, 0, sizeof(union dpni_statistics));
+
+   PMD_INIT_FUNC_TRACE();
+
+   if (!dpni) {
+   RTE_LOG(ERR, PMD, "dpni is NULL");
+   return;
+   }
+
+   if (!stats) {
+   RTE_LOG(ERR, PMD, "stats is NULL");
+   return;
+   }
+
+   /*Get Counters from page_0*/
+   retcode = dpni_get_statistics(dpni, CMD_PRI_LOW, priv->token,
+ page0, &value);
+   if (retcode)
+   goto err;
+
+   stats->ipackets = value.page_0.ingress_all_frames;
+   stats->ibytes = value.page_0.ingress_all_bytes;
+
+   /*Get Counters from page_1*/
+   retcode = dpni_get_statistics(dpni, CMD_PRI_LOW, priv->token,
+ page1, &value);
+   if (retcode)
+   goto err;
+
+   stats->opackets = value.page_1.egress_all_frames;
+   stats->obytes = value.page_1.egress_all_bytes;
+
+   /*Get Counters from page_2*/
+   retcode = dpni_get_statistics(dpni, CMD_PRI_LOW, priv->token,
+ page2, &value);
+   if (retcode)
+   goto err;
+
+   stats->ierrors = value.page_2.ingress_discarded_frames;
+   stats->oerrors = value.page_2.egress_discarded_frames;
+   stats->imissed = value.page_2.ingress_nobuffer_discards;
+
+   return;
+
+err:
+   RTE_LOG(ERR, PMD, "Operation not completed:Error Code = %d\n", retcode);
+   return;
+};
+
+static
+void dpaa2_dev_stats_reset(struct rte_eth_dev *dev)
+{
+   struct dpaa2_dev_priv *priv = dev->data->dev_private;
+   struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
+   int32_t  retcode;
+
+   PMD_INIT_FUNC_TRACE();
+
+   if (dpni == NULL) {
+   RTE_LOG(ERR, PMD, "dpni is NULL");
+   return;
+   }
+
+   retcode =  dpni_reset_statistics(dpni, CMD_PRI_LOW, priv->token);
+   if (retcode)
+   goto error;
+
+   return;
+
+error:
+   RTE_LOG(ERR, PMD, "Operation not completed:Error Code = %d\n", retcode);
+   return;
+};
+
 /* return 0 means link status changed, -1 means not changed */
 static int
 dpaa2_dev_link_update(struct rte_eth_dev *dev,
@@ -645,6 +729,8 @@
.promiscuous_enable   = dpaa2_dev_promiscuous_enable,
.promiscuous_disable  = dpaa2_dev_promiscuous_disable,
.link_update   = dpaa2_dev_link_update,
+   .stats_get = dpaa2_dev_stats_get,
+   .stats_reset   = dpaa2_dev_stats_reset,
.dev_infos_get = dpaa2_dev_info_get,
.dev_supported_ptypes_get = dpaa2_supported_ptypes_get,
.mtu_set   = dpaa2_dev_mtu_set,
-- 
1.9.1

[dpdk-dev] [PATCHv6 29/33] net/dpaa2: enable stashing for LS2088A devices

[Hemant Agrawal]

As the hardware determines which core will process which packet,
performance is boosted by direct cache warming/stashing as well
as by providing biasing for core-to-flow affinity, which ensures
that flow-specific data structures can remain in the core’s cache.

This patch enables the one cache line data stashing for packet
annotation data and packet context

Signed-off-by: Hemant Agrawal 
---
 drivers/net/dpaa2/dpaa2_ethdev.c | 11 +++
 1 file changed, 11 insertions(+)

diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index eed0136..6609a9b 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -277,6 +277,17 @@
options = options | DPNI_QUEUE_OPT_USER_CTX;
cfg.user_context = (uint64_t)(dpaa2_q);
 
+   /*if ls2088 or rev2 device, enable the stashing */
+   if ((qbman_get_version() & 0x) > QMAN_REV_4000) {
+   options |= DPNI_QUEUE_OPT_FLC;
+   cfg.flc.stash_control = true;
+   cfg.flc.value &= 0xFFC0;
+   /* 00 00 00 - last 6 bit represent annotation, context stashing,
+* data stashing setting 01 01 00 (0x14) to enable
+* 1 line annotation, 1 line context
+*/
+   cfg.flc.value |= 0x14;
+   }
ret = dpni_set_queue(dpni, CMD_PRI_LOW, priv->token, DPNI_QUEUE_RX,
 dpaa2_q->tc_index, flow_id, options, &cfg);
if (ret) {
-- 
1.9.1

[dpdk-dev] [PATCHv6 32/33] bus/fslmc: add support for dmamap to ARM SMMU

[Hemant Agrawal]

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/fslmc_vfio.c  | 96 +
 drivers/bus/fslmc/fslmc_vfio.h  |  1 +
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  1 +
 drivers/net/dpaa2/dpaa2_ethdev.c|  2 +
 4 files changed, 100 insertions(+)

diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index fc017fc..5f1d1b7 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -76,8 +76,10 @@
 static struct fslmc_vfio_group vfio_groups[VFIO_MAX_GRP];
 static struct fslmc_vfio_container vfio_containers[VFIO_MAX_CONTAINERS];
 static int container_device_fd;
+static uint32_t *msi_intr_vaddr;
 void *(*rte_mcp_ptr_list);
 static uint32_t mcp_id;
+static int is_dma_done;
 
 static int vfio_connect_container(struct fslmc_vfio_group *vfio_group)
 {
@@ -147,6 +149,35 @@ static int vfio_connect_container(struct fslmc_vfio_group 
*vfio_group)
return 0;
 }
 
+static int vfio_map_irq_region(struct fslmc_vfio_group *group)
+{
+   int ret;
+   unsigned long *vaddr = NULL;
+   struct vfio_iommu_type1_dma_map map = {
+   .argsz = sizeof(map),
+   .flags = VFIO_DMA_MAP_FLAG_READ | VFIO_DMA_MAP_FLAG_WRITE,
+   .vaddr = 0x603,
+   .iova = 0x603,
+   .size = 0x1000,
+   };
+
+   vaddr = (unsigned long *)mmap(NULL, 0x1000, PROT_WRITE |
+   PROT_READ, MAP_SHARED, container_device_fd, 0x603);
+   if (vaddr == MAP_FAILED) {
+   FSLMC_VFIO_LOG(ERR, "Unable to map region (errno = %d)", errno);
+   return -errno;
+   }
+
+   msi_intr_vaddr = (uint32_t *)((char *)(vaddr) + 64);
+   map.vaddr = (unsigned long)vaddr;
+   ret = ioctl(group->container->fd, VFIO_IOMMU_MAP_DMA, &map);
+   if (ret == 0)
+   return 0;
+
+   FSLMC_VFIO_LOG(ERR, "VFIO_IOMMU_MAP_DMA fails (errno = %d)", errno);
+   return -errno;
+}
+
 int vfio_dmamap_mem_region(uint64_t vaddr,
   uint64_t iova,
   uint64_t size)
@@ -170,6 +201,71 @@ int vfio_dmamap_mem_region(uint64_t vaddr,
return 0;
 }
 
+int rte_fslmc_vfio_dmamap(void)
+{
+   int ret;
+   struct fslmc_vfio_group *group;
+   struct vfio_iommu_type1_dma_map dma_map = {
+   .argsz = sizeof(struct vfio_iommu_type1_dma_map),
+   .flags = VFIO_DMA_MAP_FLAG_READ | VFIO_DMA_MAP_FLAG_WRITE,
+   };
+
+   int i;
+   const struct rte_memseg *memseg;
+
+   if (is_dma_done)
+   return 0;
+   is_dma_done = 1;
+
+   for (i = 0; i < RTE_MAX_MEMSEG; i++) {
+   memseg = rte_eal_get_physmem_layout();
+   if (memseg == NULL) {
+   FSLMC_VFIO_LOG(ERR, "Cannot get physical layout.");
+   return -ENODEV;
+   }
+
+   if (memseg[i].addr == NULL && memseg[i].len == 0)
+   break;
+
+   dma_map.size = memseg[i].len;
+   dma_map.vaddr = memseg[i].addr_64;
+#ifdef RTE_LIBRTE_DPAA2_USE_PHYS_IOVA
+   dma_map.iova = memseg[i].phys_addr;
+#else
+   dma_map.iova = dma_map.vaddr;
+#endif
+
+   /* SET DMA MAP for IOMMU */
+   group = &vfio_groups[0];
+
+   if (!group->container) {
+   FSLMC_VFIO_LOG(ERR, "Container is not connected ");
+   return -1;
+   }
+
+   FSLMC_VFIO_LOG(DEBUG, "-->Initial SHM Virtual ADDR %llX",
+dma_map.vaddr);
+   FSLMC_VFIO_LOG(DEBUG, "-> DMA size 0x%llX\n", dma_map.size);
+   ret = ioctl(group->container->fd, VFIO_IOMMU_MAP_DMA,
+   &dma_map);
+   if (ret) {
+   FSLMC_VFIO_LOG(ERR, "VFIO_IOMMU_MAP_DMA API"
+  "(errno = %d)", errno);
+   return ret;
+   }
+   FSLMC_VFIO_LOG(DEBUG, "-> dma_map.vaddr = 0x%llX",
+dma_map.vaddr);
+   }
+
+   /* TODO - This is a W.A. as VFIO currently does not add the mapping of
+* the interrupt region to SMMU. This should be removed once the
+* support is added in the Kernel.
+*/
+   vfio_map_irq_region(group);
+
+   return 0;
+}
+
 static int64_t vfio_map_mcp_obj(struct fslmc_vfio_group *group, char *mcp_obj)
 {
int64_t v_addr = (int64_t)MAP_FAILED;
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index 80c6869..53dd0b7 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -70,6 +70,7 @@ int vfio_dmamap_mem_region(
 
 int fslmc_vfio_setup_group(void);
 int fslmc_vfio_process_group(void);
+int rte_fslmc_vfio_dmamap(void);
 
 /* create dpio device */
 int dpaa2_create_dpio_device(struct fslmc_vfio_device

[dpdk-dev] [PATCHv6 31/33] net/dpaa2: enabling the use of physical addresses

[Hemant Agrawal]

DPAA2 - ARM support both physical and virtual addressing.
This patch enables the compile time usages of physical
address instead of virtual address.

The current usages are also set to default as Physical
Address.

Signed-off-by: Hemant Agrawal 
---
 config/common_base|  1 +
 config/defconfig_arm64-dpaa2-linuxapp-gcc |  1 +
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h   | 66 +++
 drivers/net/dpaa2/base/dpaa2_hw_dpni.c|  4 +-
 drivers/net/dpaa2/dpaa2_rxtx.c| 16 
 drivers/pool/dpaa2/dpaa2_hw_mempool.c | 19 +++--
 6 files changed, 95 insertions(+), 12 deletions(-)

diff --git a/config/common_base b/config/common_base
index dd3de11..6f1787a 100644
--- a/config/common_base
+++ b/config/common_base
@@ -291,6 +291,7 @@ CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_MBOX=n
 # Compile Support Libraries for NXP DPAA2
 #
 CONFIG_RTE_LIBRTE_DPAA2_POOL=n
+CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=y
 
 #
 # Compile NXP DPAA2 FSL-MC Bus
diff --git a/config/defconfig_arm64-dpaa2-linuxapp-gcc 
b/config/defconfig_arm64-dpaa2-linuxapp-gcc
index 3cdb31b..29a56c7 100644
--- a/config/defconfig_arm64-dpaa2-linuxapp-gcc
+++ b/config/defconfig_arm64-dpaa2-linuxapp-gcc
@@ -49,6 +49,7 @@ CONFIG_RTE_PKTMBUF_HEADROOM=256
 #
 CONFIG_RTE_LIBRTE_DPAA2_POOL=n
 CONFIG_RTE_MBUF_DEFAULT_MEMPOOL_OPS="dpaa2"
+CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=y
 
 #
 # Compile NXP DPAA2 FSL-MC Bus
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index c26360d3..ad8a22f 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -175,6 +175,72 @@ struct qbman_fle {
  */
 #define DPAA2_EQ_RESP_ALWAYS   1
 
+#ifdef RTE_LIBRTE_DPAA2_USE_PHYS_IOVA
+static void *dpaa2_mem_ptov(phys_addr_t paddr) __attribute__((unused));
+/* todo - this is costly, need to write a fast coversion routine */
+static void *dpaa2_mem_ptov(phys_addr_t paddr)
+{
+   const struct rte_memseg *memseg = rte_eal_get_physmem_layout();
+   int i;
+
+   for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
+   if (paddr >= memseg[i].phys_addr &&
+  (char *)paddr < (char *)memseg[i].phys_addr + memseg[i].len)
+   return (void *)(memseg[i].addr_64
+   + (paddr - memseg[i].phys_addr));
+   }
+   return NULL;
+}
+
+static phys_addr_t dpaa2_mem_vtop(uint64_t vaddr) __attribute__((unused));
+static phys_addr_t dpaa2_mem_vtop(uint64_t vaddr)
+{
+   const struct rte_memseg *memseg = rte_eal_get_physmem_layout();
+   int i;
+
+   for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
+   if (vaddr >= memseg[i].addr_64 &&
+   vaddr < memseg[i].addr_64 + memseg[i].len)
+   return memseg[i].phys_addr
+   + (vaddr - memseg[i].addr_64);
+   }
+   return (phys_addr_t)(NULL);
+}
+
+/**
+ * When we are using Physical addresses as IO Virtual Addresses,
+ * Need to call conversion routines dpaa2_mem_vtop & dpaa2_mem_ptov
+ * whereever required.
+ * These routines are called with help of below MACRO's
+ */
+
+#define DPAA2_MBUF_VADDR_TO_IOVA(mbuf) ((mbuf)->buf_physaddr)
+
+/**
+ * macro to convert Virtual address to IOVA
+ */
+#define DPAA2_VADDR_TO_IOVA(_vaddr) dpaa2_mem_vtop((uint64_t)(_vaddr))
+
+/**
+ * macro to convert IOVA to Virtual address
+ */
+#define DPAA2_IOVA_TO_VADDR(_iova) dpaa2_mem_ptov((phys_addr_t)(_iova))
+
+/**
+ * macro to convert modify the memory containing IOVA to Virtual address
+ */
+#define DPAA2_MODIFY_IOVA_TO_VADDR(_mem, _type) \
+   {_mem = (_type)(dpaa2_mem_ptov((phys_addr_t)(_mem))); }
+
+#else  /* RTE_LIBRTE_DPAA2_USE_PHYS_IOVA */
+
+#define DPAA2_MBUF_VADDR_TO_IOVA(mbuf) ((mbuf)->buf_addr)
+#define DPAA2_VADDR_TO_IOVA(_vaddr) (_vaddr)
+#define DPAA2_IOVA_TO_VADDR(_iova) (_iova)
+#define DPAA2_MODIFY_IOVA_TO_VADDR(_mem, _type)
+
+#endif /* RTE_LIBRTE_DPAA2_USE_PHYS_IOVA */
+
 struct dpaa2_dpbp_dev *dpaa2_alloc_dpbp_dev(void);
 void dpaa2_free_dpbp_dev(struct dpaa2_dpbp_dev *dpbp);
 
diff --git a/drivers/net/dpaa2/base/dpaa2_hw_dpni.c 
b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
index 08f53b3..3dc60cc 100644
--- a/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
+++ b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
@@ -76,7 +76,7 @@
memset(&tc_cfg, 0, sizeof(struct dpni_rx_tc_dist_cfg));
 
dpaa2_distset_to_dpkg_profile_cfg(req_dist_set, &kg_cfg);
-   tc_cfg.key_cfg_iova = (uint64_t)(p_params);
+   tc_cfg.key_cfg_iova = (uint64_t)(DPAA2_VADDR_TO_IOVA(p_params));
tc_cfg.dist_size = eth_dev->data->nb_rx_queues;
tc_cfg.dist_mode = DPNI_DIST_MODE_HASH;
 
@@ -119,7 +119,7 @@ int dpaa2_remove_flow_dist(
memset(p_params, 0, DIST_PARAM_IOVA_SIZE);
memset(&tc_cfg, 0, sizeof(struct dpni_rx_tc_dist_cfg));
 
-   tc_cfg.key_cfg_iova = (uint64_t)(p_params);
+   tc_cfg.key_cfg_iova 

Re: [dpdk-dev] [dpdk-stable] [PATCH] net/mlx5: fix memory leak when parsing device params

[Ferruh Yigit]

On 1/22/2017 8:24 AM, Shahaf Shuler wrote:
> in case of an error argument list is not freed.
> 
> Fixes: e72dd09b614e ("net/mlx5: add support for configuration through kvargs")
> CC: sta...@dpdk.org
> 
> Signed-off-by: Shahaf Shuler 


Applied to dpdk-next-net/master, thanks.

[dpdk-dev] [PATCHv6 33/33] drivers/common/dpaa2: frame queue based dq storage alloc

[Hemant Agrawal]

This patch adds generic functions for allowing dq storage
for the frame queues.
As the frame queues are common resource for different drivers
this is helpful.

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c| 32 +
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h|  7 +++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  2 ++
 drivers/net/dpaa2/dpaa2_ethdev.c|  8 
 4 files changed, 45 insertions(+), 4 deletions(-)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
index bd1f643..c80d6c5 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -407,3 +407,35 @@ static inline struct dpaa2_dpio_dev 
*dpaa2_get_qbman_swp(void)
 
return 0;
 }
+
+void
+dpaa2_free_dq_storage(struct queue_storage_info_t *q_storage)
+{
+   int i = 0;
+
+   for (i = 0; i < NUM_DQS_PER_QUEUE; i++) {
+   if (q_storage->dq_storage[i])
+   rte_free(q_storage->dq_storage[i]);
+   }
+}
+
+int
+dpaa2_alloc_dq_storage(struct queue_storage_info_t *q_storage)
+{
+   int i = 0;
+
+   for (i = 0; i < NUM_DQS_PER_QUEUE; i++) {
+   q_storage->dq_storage[i] = rte_malloc(NULL,
+   DPAA2_DQRR_RING_SIZE * sizeof(struct qbman_result),
+   RTE_CACHE_LINE_SIZE);
+   if (!q_storage->dq_storage[i])
+   goto fail;
+   }
+   return 0;
+fail:
+   i -= 1;
+   while (i >= 0)
+   rte_free(q_storage->dq_storage[i]);
+
+   return -1;
+}
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
index b1a1b8f..f2e1168 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
@@ -59,5 +59,12 @@ struct dpaa2_io_portal_t {
 /* Affine additional DPIO portal to current crypto processing thread */
 int dpaa2_affine_qbman_swp_sec(void);
 
+/* allocate memory for FQ - dq storage */
+int
+dpaa2_alloc_dq_storage(struct queue_storage_info_t *q_storage);
+
+/* free memory for FQ- dq storage */
+void
+dpaa2_free_dq_storage(struct queue_storage_info_t *q_storage);
 
 #endif /* _DPAA2_HW_DPIO_H_ */
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index 505873a..4298d77 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -4,7 +4,9 @@ DPDK_17.02 {
 dpaa2_affine_qbman_swp;
 dpaa2_affine_qbman_swp_sec;
 dpaa2_alloc_dpbp_dev;
+dpaa2_alloc_dq_storage;
 dpaa2_free_dpbp_dev;
+dpaa2_free_dq_storage;
 dpbp_disable;
 dpbp_enable;
 dpbp_get_attributes;
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index cbfbc7a..67eb34d 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -49,6 +49,7 @@
 #include 
 #include 
 #include 
+#include 
 
 #include "dpaa2_ethdev.h"
 
@@ -169,9 +170,8 @@
 
memset(dpaa2_q->q_storage, 0,
   sizeof(struct queue_storage_info_t));
-   dpaa2_q->q_storage->dq_storage[0] = rte_malloc(NULL,
-   DPAA2_DQRR_RING_SIZE * sizeof(struct qbman_result),
-   RTE_CACHE_LINE_SIZE);
+   if (dpaa2_alloc_dq_storage(dpaa2_q->q_storage))
+   goto fail;
}
 
for (i = 0; i < priv->nb_tx_queues; i++) {
@@ -195,7 +195,7 @@
mc_q = priv->rx_vq[0];
while (i >= 0) {
dpaa2_q = (struct dpaa2_queue *)priv->rx_vq[i];
-   rte_free(dpaa2_q->q_storage->dq_storage[0]);
+   dpaa2_free_dq_storage(dpaa2_q->q_storage);
rte_free(dpaa2_q->q_storage);
priv->rx_vq[i--] = NULL;
}
-- 
1.9.1

[dpdk-dev] [PATCH] buildtools: allow symlinks within a source directory

[Bruce Richardson]

When creating the symlinks for header files to the include folder, the
relpath script dereferenced all symlinks. This made it impossible to
have file A.h renamed to B.h and then symlinked back to its original
name. This is useful to be able to do when refactoring or reworking
a library. Change this so that we just use the dirname of the path from
readlink, we can use the basename as it was originally, even if it was a
symlink.

Signed-off-by: Bruce Richardson 
---
 buildtools/relpath.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/buildtools/relpath.sh b/buildtools/relpath.sh
index 4ff4671..139b781 100755
--- a/buildtools/relpath.sh
+++ b/buildtools/relpath.sh
@@ -44,7 +44,7 @@ if [ $# -ne 2 ]; then
 fi
 
 # get the real absolute path, derefencing symlinks
-ABS1=$(readlink -f $1)
+ABS1="$(dirname $(readlink -f $1))/$(basename $1)"
 ABS2=$(readlink -f $2)
 
 # remove leading slash
-- 
2.9.3

[dpdk-dev] [PATCH] cryptodev: crypto PMD functions incorrectly inlined

[Declan Doherty]

rte_cryptodev_pmd_get_dev, rte_cryptodev_pmd_get_named_dev, 
rte_cryptodev_pmd_is_valid_dev were incorrectly marked as inline and
therefore not useable from crypto PMDs when built as shared
libraries as they accessed the global rte_cryptodev_globals device
structure.

Fixes: d11b0f30 ("cryptodev: introduce API and framework for crypto
devices")

Signed-off-by: Declan Doherty 
---
 lib/librte_cryptodev/rte_cryptodev.c   | 42 
 lib/librte_cryptodev/rte_cryptodev_pmd.h   | 44 --
 lib/librte_cryptodev/rte_cryptodev_version.map |  3 ++
 3 files changed, 51 insertions(+), 38 deletions(-)

diff --git a/lib/librte_cryptodev/rte_cryptodev.c 
b/lib/librte_cryptodev/rte_cryptodev.c
index 6a51eec..42707cb 100644
--- a/lib/librte_cryptodev/rte_cryptodev.c
+++ b/lib/librte_cryptodev/rte_cryptodev.c
@@ -255,6 +255,48 @@ rte_cryptodev_create_vdev(const char *name, const char 
*args)
return rte_eal_vdev_init(name, args);
 }
 
+struct rte_cryptodev *
+rte_cryptodev_pmd_get_dev(uint8_t dev_id)
+{
+   return &rte_cryptodev_globals->devs[dev_id];
+}
+
+struct rte_cryptodev *
+rte_cryptodev_pmd_get_named_dev(const char *name)
+{
+   struct rte_cryptodev *dev;
+   unsigned i;
+
+   if (name == NULL)
+   return NULL;
+
+   for (i = 0; i < rte_cryptodev_globals->max_devs; i++) {
+   dev = &rte_cryptodev_globals->devs[i];
+
+   if ((dev->attached == RTE_CRYPTODEV_ATTACHED) &&
+   (strcmp(dev->data->name, name) == 0))
+   return dev;
+   }
+
+   return NULL;
+}
+
+unsigned
+rte_cryptodev_pmd_is_valid_dev(uint8_t dev_id)
+{
+   struct rte_cryptodev *dev = NULL;
+
+   if (dev_id >= rte_cryptodev_globals->nb_devs)
+   return 0;
+
+   dev = rte_cryptodev_pmd_get_dev(dev_id);
+   if (dev->attached != RTE_CRYPTODEV_ATTACHED)
+   return 0;
+   else
+   return 1;
+}
+
+
 int
 rte_cryptodev_get_dev_id(const char *name)
 {
diff --git a/lib/librte_cryptodev/rte_cryptodev_pmd.h 
b/lib/librte_cryptodev/rte_cryptodev_pmd.h
index aabef41..b6dc32c 100644
--- a/lib/librte_cryptodev/rte_cryptodev_pmd.h
+++ b/lib/librte_cryptodev/rte_cryptodev_pmd.h
@@ -160,11 +160,8 @@ extern struct rte_cryptodev_global *rte_cryptodev_globals;
  * @return
  *   - The rte_cryptodev structure pointer for the given device ID.
  */
-static inline struct rte_cryptodev *
-rte_cryptodev_pmd_get_dev(uint8_t dev_id)
-{
-   return &rte_cryptodev_globals->devs[dev_id];
-}
+struct rte_cryptodev *
+rte_cryptodev_pmd_get_dev(uint8_t dev_id);
 
 /**
  * Get the rte_cryptodev structure device pointer for the named device.
@@ -174,25 +171,8 @@ rte_cryptodev_pmd_get_dev(uint8_t dev_id)
  * @return
  *   - The rte_cryptodev structure pointer for the given device ID.
  */
-static inline struct rte_cryptodev *
-rte_cryptodev_pmd_get_named_dev(const char *name)
-{
-   struct rte_cryptodev *dev;
-   unsigned i;
-
-   if (name == NULL)
-   return NULL;
-
-   for (i = 0; i < rte_cryptodev_globals->max_devs; i++) {
-   dev = &rte_cryptodev_globals->devs[i];
-
-   if ((dev->attached == RTE_CRYPTODEV_ATTACHED) &&
-   (strcmp(dev->data->name, name) == 0))
-   return dev;
-   }
-
-   return NULL;
-}
+struct rte_cryptodev *
+rte_cryptodev_pmd_get_named_dev(const char *name);
 
 /**
  * Validate if the crypto device index is valid attached crypto device.
@@ -202,20 +182,8 @@ rte_cryptodev_pmd_get_named_dev(const char *name)
  * @return
  *   - If the device index is valid (1) or not (0).
  */
-static inline unsigned
-rte_cryptodev_pmd_is_valid_dev(uint8_t dev_id)
-{
-   struct rte_cryptodev *dev = NULL;
-
-   if (dev_id >= rte_cryptodev_globals->nb_devs)
-   return 0;
-
-   dev = rte_cryptodev_pmd_get_dev(dev_id);
-   if (dev->attached != RTE_CRYPTODEV_ATTACHED)
-   return 0;
-   else
-   return 1;
-}
+unsigned
+rte_cryptodev_pmd_is_valid_dev(uint8_t dev_id);
 
 /**
  * The pool of rte_cryptodev structures.
diff --git a/lib/librte_cryptodev/rte_cryptodev_version.map 
b/lib/librte_cryptodev/rte_cryptodev_version.map
index c581eea..a92df62 100644
--- a/lib/librte_cryptodev/rte_cryptodev_version.map
+++ b/lib/librte_cryptodev/rte_cryptodev_version.map
@@ -51,5 +51,8 @@ DPDK_17.02 {
global:
 
rte_cryptodev_pmd_create_dev_name;
+   rte_cryptodev_pmd_get_dev;
+   rte_cryptodev_pmd_get_named_dev;
+   rte_cryptodev_pmd_is_valid_dev;
 
 } DPDK_16.11;
-- 
2.9.3

Re: [dpdk-dev] [PATCH v1] net/ixgbe: add more check in n-tuple filter

[Ferruh Yigit]

On 1/23/2017 3:32 AM, Wei Zhao wrote:
> Add more check on the mask of src_addr, dst_addr and next_proto_id
> in n-tuple filter rule pattern parser.If do not add such check,
> it maybe cause error in pattern parser.
> 
> Signed-off-by: zhao wei 
> ---
>  drivers/net/ixgbe/ixgbe_flow.c | 5 -
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
> index 82aceed..8f78eee 100644
> --- a/drivers/net/ixgbe/ixgbe_flow.c
> +++ b/drivers/net/ixgbe/ixgbe_flow.c
> @@ -331,7 +331,10 @@ cons_parse_ntuple_filter(const struct rte_flow_attr 
> *attr,
>   ipv4_mask->hdr.packet_id ||
>   ipv4_mask->hdr.fragment_offset ||
>   ipv4_mask->hdr.time_to_live ||
> - ipv4_mask->hdr.hdr_checksum) {
> + ipv4_mask->hdr.hdr_checksum ||
> + !ipv4_mask->hdr.next_proto_id ||
> + !ipv4_mask->hdr.dst_addr ||
> + !ipv4_mask->hdr.src_addr ||) {

The '||' at the end seems extra.

Also can you please confirm your sign off?

>   rte_flow_error_set(error,
>   EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
>   item, "Not supported by ntuple filter");
> 

Re: [dpdk-dev] [PATCH] cryptodev: crypto PMD functions incorrectly inlined

[Zhang, Roy Fan]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Declan Doherty
> Sent: Monday, January 23, 2017 12:19 PM
> To: dev@dpdk.org
> Cc: De Lara Guarch, Pablo ;
> sta...@dpdk.org; Doherty, Declan 
> Subject: [dpdk-dev] [PATCH] cryptodev: crypto PMD functions incorrectly
> inlined
> 
> rte_cryptodev_pmd_get_dev, rte_cryptodev_pmd_get_named_dev,
> rte_cryptodev_pmd_is_valid_dev were incorrectly marked as inline and
> therefore not useable from crypto PMDs when built as shared libraries as
> they accessed the global rte_cryptodev_globals device structure.
> 
> Fixes: d11b0f30 ("cryptodev: introduce API and framework for crypto
> devices")
> 
> Signed-off-by: Declan Doherty 
> ---
>  lib/librte_cryptodev/rte_cryptodev.c   | 42
> 
>  lib/librte_cryptodev/rte_cryptodev_pmd.h   | 44 
> --
>  lib/librte_cryptodev/rte_cryptodev_version.map |  3 ++
>  3 files changed, 51 insertions(+), 38 deletions(-)
> 
> diff --git a/lib/librte_cryptodev/rte_cryptodev.c
> b/lib/librte_cryptodev/rte_cryptodev.c
> index 6a51eec..42707cb 100644
> --- a/lib/librte_cryptodev/rte_cryptodev.c
> +++ b/lib/librte_cryptodev/rte_cryptodev.c
> @@ -255,6 +255,48 @@ rte_cryptodev_create_vdev(const char *name,
> const char *args)
>   return rte_eal_vdev_init(name, args);
>  }
> 
> +struct rte_cryptodev *
> +rte_cryptodev_pmd_get_dev(uint8_t dev_id) {
> + return &rte_cryptodev_globals->devs[dev_id];
> +}
> +
> +struct rte_cryptodev *
> +rte_cryptodev_pmd_get_named_dev(const char *name) {
> + struct rte_cryptodev *dev;
> + unsigned i;
> +
> + if (name == NULL)
> + return NULL;
> +
> + for (i = 0; i < rte_cryptodev_globals->max_devs; i++) {
> + dev = &rte_cryptodev_globals->devs[i];
> +
> + if ((dev->attached == RTE_CRYPTODEV_ATTACHED) &&
> + (strcmp(dev->data->name, name) == 0))
> + return dev;
> + }
> +
> + return NULL;
> +}
> +
> +unsigned
> +rte_cryptodev_pmd_is_valid_dev(uint8_t dev_id) {
> + struct rte_cryptodev *dev = NULL;
> +
> + if (dev_id >= rte_cryptodev_globals->nb_devs)
> + return 0;
> +
> + dev = rte_cryptodev_pmd_get_dev(dev_id);
> + if (dev->attached != RTE_CRYPTODEV_ATTACHED)
> + return 0;
> + else
> + return 1;
> +}
> +
> +
>  int
>  rte_cryptodev_get_dev_id(const char *name)  { diff --git
> a/lib/librte_cryptodev/rte_cryptodev_pmd.h
> b/lib/librte_cryptodev/rte_cryptodev_pmd.h
> index aabef41..b6dc32c 100644
> --- a/lib/librte_cryptodev/rte_cryptodev_pmd.h
> +++ b/lib/librte_cryptodev/rte_cryptodev_pmd.h
> @@ -160,11 +160,8 @@ extern struct rte_cryptodev_global
> *rte_cryptodev_globals;
>   * @return
>   *   - The rte_cryptodev structure pointer for the given device ID.
>   */
> -static inline struct rte_cryptodev *
> -rte_cryptodev_pmd_get_dev(uint8_t dev_id) -{
> - return &rte_cryptodev_globals->devs[dev_id];
> -}
> +struct rte_cryptodev *
> +rte_cryptodev_pmd_get_dev(uint8_t dev_id);
> 
>  /**
>   * Get the rte_cryptodev structure device pointer for the named device.
> @@ -174,25 +171,8 @@ rte_cryptodev_pmd_get_dev(uint8_t dev_id)
>   * @return
>   *   - The rte_cryptodev structure pointer for the given device ID.
>   */
> -static inline struct rte_cryptodev *
> -rte_cryptodev_pmd_get_named_dev(const char *name) -{
> - struct rte_cryptodev *dev;
> - unsigned i;
> -
> - if (name == NULL)
> - return NULL;
> -
> - for (i = 0; i < rte_cryptodev_globals->max_devs; i++) {
> - dev = &rte_cryptodev_globals->devs[i];
> -
> - if ((dev->attached == RTE_CRYPTODEV_ATTACHED) &&
> - (strcmp(dev->data->name, name) == 0))
> - return dev;
> - }
> -
> - return NULL;
> -}
> +struct rte_cryptodev *
> +rte_cryptodev_pmd_get_named_dev(const char *name);
> 
>  /**
>   * Validate if the crypto device index is valid attached crypto device.
> @@ -202,20 +182,8 @@ rte_cryptodev_pmd_get_named_dev(const char
> *name)
>   * @return
>   *   - If the device index is valid (1) or not (0).
>   */
> -static inline unsigned
> -rte_cryptodev_pmd_is_valid_dev(uint8_t dev_id) -{
> - struct rte_cryptodev *dev = NULL;
> -
> - if (dev_id >= rte_cryptodev_globals->nb_devs)
> - return 0;
> -
> - dev = rte_cryptodev_pmd_get_dev(dev_id);
> - if (dev->attached != RTE_CRYPTODEV_ATTACHED)
> - return 0;
> - else
> - return 1;
> -}
> +unsigned
> +rte_cryptodev_pmd_is_valid_dev(uint8_t dev_id);
> 
>  /**
>   * The pool of rte_cryptodev structures.
> diff --git a/lib/librte_cryptodev/rte_cryptodev_version.map
> b/lib/librte_cryptodev/rte_cryptodev_version.map
> index c581eea..a92df62 100644
> --- a/lib/librte_cryptodev/rte_cryptodev_version.map
> +++ b/lib/librte_cryptodev/rte_cryptodev_version.map
> @@ -51,5 +51,8 @@ DPDK_17.02 {
>   global:
> 
>   

Re: [dpdk-dev] [PATCH v5 1/6] lib: distributor performance enhancements

[Bruce Richardson]

On Fri, Jan 20, 2017 at 09:18:48AM +, David Hunt wrote:
> Now sends bursts of up to 8 mbufs to each worker, and tracks
> the in-flight flow-ids (atomic scheduling)
> 
> New file with a new api, similar to the old API except with _burst
> at the end of the function names. This is to preserve the original
> API (and code) for backward compatibility.
> 
> It uses a similar handshake mechanism to the previous version of
> the library, in that bits are used to indicate when packets are ready
> to be sent to a worker and ready to be returned from a worker. One main
> difference is that instead of sending one packet in a cache line, it
> makes use of the 7 free spaces in the same cache line in order to send up to
> 8 packets at a time to/from a worker.
> 
> The flow matching algorithm has had significant re-work, and now keeps
> an array of inflight flows and an array of backlog flows, and matches
> incoming flows to the inflight/backlog flows of all workers so
> that flow pinning to workers can be maintained.
> 
> Signed-off-by: David Hunt 
> ---

I still see some issues reported here by sanity check scripts.

/Bruce

--- /dev/null   2017-01-10 10:26:01.206201474 +
+++ /tmp/doc-check/doc.txt  2017-01-23 12:23:01.748870247 +
@@ -0,0 +1,6 @@
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_burst.h:187:
warning: argument 'mbuf' of command @param is not found in the argument
list of rte_distributor_return_pkt_burst(struct rte_distributor_burst
*d, unsigned int worker_id, struct rte_mbuf **oldpkt, int num)
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_burst.h:199:
warning: The following parameters of
rte_distributor_return_pkt_burst(struct rte_distributor_burst *d,
unsigned int worker_id, struct rte_mbuf **oldpkt, int num) are not
documented:
+  parameter 'oldpkt'
+  parameter 'num'
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_priv.h:73:
warning: Found unknown command `\in_flight_bitmask'
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_priv.h:73:
warning: Found unknown command `\rte_distributor_process'
 Error with doc check 
Line too long:
makes use of the 7 free spaces in the same cache line in order
to send up to

### lib: distributor performance enhancements

WARNING:COMMIT_LOG_LONG_LINE: Possible unwrapped commit description
(prefer a maximum 75 chars per line)
#17:
makes use of the 7 free spaces in the same cache line in order to send
up to

total: 0 errors, 1 warnings, 1131 lines checked

0/1 valid patch

Re: [dpdk-dev] [PATCH 1/5] net/mlx5: last WQE no room inline

[Ferruh Yigit]

On 1/8/2017 3:41 PM, Elad Persiko wrote:
> Prior to this patch, when sending a packet and the following
> conditions were reached:
>   1. last working queue element is used.
>   2. inline was requested by the user
>   3. no room for inline packet.
> then the inline request was ignored and the packet was sent
> by pointer completely.
> 
> This patch handles this scenario. In this case the last
> work queue element is turned to be a null work queue element and
> the packet is being sent after the wrap around.
> 
> Signed-off-by: Elad Persiko 

It looks like this series missing driver maintainer CC'ed.

Cc: Adrien Mazarguil 

Re: [dpdk-dev] [PATCH v2] doc: add tso capabilities feature for mlx5

[Ferruh Yigit]

On 1/10/2017 9:18 AM, Elad Persiko wrote:
> 
> Thanks,
> The feature is not supported on MLX4. I will fix it on V2

The new version of this patch will be squashed to "net/mlx5: last WQE no
room inline" patchset, right?

> 
> -Original Message-
> From: Ferruh Yigit [mailto:ferruh.yi...@intel.com]
> Sent: Monday, January 9, 2017 2:31 PM
> To: Elad Persiko ; dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] doc: add tso capabilities feature for mlx5
> 
> On 1/8/2017 4:23 PM, Elad Persiko wrote:
>> Feature implemented at:
>> commit b007e98ccda9 ("net/mlx5: implement TSO data path") commit 
>> 085c4137280a ("net/mlx5: support TSO in control plane")
>>
>> Signed-off-by: Elad Persiko 
>> ---
>>  doc/guides/nics/features/mlx4.ini | 1 + 
>> doc/guides/nics/features/mlx5.ini | 1 +
>>  2 files changed, 2 insertions(+)
>>
>> diff --git a/doc/guides/nics/features/mlx4.ini
>> b/doc/guides/nics/features/mlx4.ini
>> index c9828f7..d74b9dd 100644
>> --- a/doc/guides/nics/features/mlx4.ini
>> +++ b/doc/guides/nics/features/mlx4.ini
>> @@ -10,6 +10,7 @@ Queue start/stop = Y
>>  MTU update   = Y
>>  Jumbo frame  = Y
>>  Scattered Rx = Y
>> +TSO  = Y
> 
> One more thing, for double check, patch updates only mlx5 files, is enabling 
> TSO for mlx4 correct?
> 
>>  Promiscuous mode = Y
>>  Allmulticast mode= Y
>>  Unicast MAC filter   = Y
> <...>
> 

Re: [dpdk-dev] [PATCH v5 5/6] examples/distributor_app: showing burst API

[Bruce Richardson]

On Fri, Jan 20, 2017 at 09:18:52AM +, David Hunt wrote:
> Signed-off-by: David Hunt 
> ---
>  examples/distributor/main.c | 509 
> ++--
>  1 file changed, 391 insertions(+), 118 deletions(-)
> 

Another minor nit from checkpatch.

/Bruce

### examples/distributor_app: showing burst API

WARNING:BLOCK_COMMENT_STYLE: Block comments should align the * on each
line
#301: FILE: examples/distributor/main.c:327:
+   /*
+   * Swap the following two lines if you want the rx
traffic

total: 0 errors, 1 warnings, 744 lines checked

0/1 valid patch

Re: [dpdk-dev] [PATCH v2] doc: add tso capabilities feature for mlx5

[Olga Shern]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Ferruh Yigit
> Sent: Monday, January 23, 2017 2:30 PM
> To: Elad Persiko ; dev@dpdk.org; Adrien Mazarguil
> 
> Subject: Re: [dpdk-dev] [PATCH v2] doc: add tso capabilities feature for mlx5
> 
> On 1/10/2017 9:18 AM, Elad Persiko wrote:
> >
> > Thanks,
> > The feature is not supported on MLX4. I will fix it on V2
> 
> The new version of this patch will be squashed to "net/mlx5: last WQE no
> room inline" patchset, right?

Yes, you are right 

Best Regards,
Olga

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Ananyev, Konstantin]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Yuanhan Liu
> Sent: Monday, January 23, 2017 11:56 AM
> To: Yigit, Ferruh 
> Cc: dev@dpdk.org; Thomas Monjalon ; Horton, Remy 
> 
> Subject: Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset
> 
> On Mon, Jan 23, 2017 at 07:40:50PM +0800, Yuanhan Liu wrote:
> > On Mon, Jan 23, 2017 at 11:32:23AM +, Ferruh Yigit wrote:
> > > On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
> > > > On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
> > >   lib/librte_ether/rte_ethdev.c | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > >  diff --git a/lib/librte_ether/rte_ethdev.c 
> > >  b/lib/librte_ether/rte_ethdev.c
> > >  index 4790faf..61f44e2 100644
> > >  --- a/lib/librte_ether/rte_ethdev.c
> > >  +++ b/lib/librte_ether/rte_ethdev.c
> > >  @@ -225,7 +225,7 @@ struct rte_eth_dev *
> > >   return NULL;
> > >   }
> > > 
> > >  -memset(&rte_eth_devices[port_id], 0, 
> > >  sizeof(*eth_dev->data));
> > >  +memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
> > >  rte_eth_dev_data));
> > > >>>
> > > >>> Not directly related to the this issue, but, after fix, this may 
> > > >>> have
> > > >>> issues with secondary process.
> > > >>>
> > > >>> There were patches sent to fix this.
> > > >>
> > > >> I mean this one:
> > > >> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > > >
> > > > d948f596fee2 ("ethdev: fix port data mismatched in multiple process
> > > > model") should have fixed it.
> > > 
> > >  Think about case, where secondary process uses a virtual PMD, which 
> > >  does
> > >  a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
> > >  device data?
> > > >>>
> > > >>> Yes, it may. However, I doubt that's the typical usage.
> > > >>
> > > >> But this is a use case, and broken now,
> > > >
> > > > I thought it was broken since the beginning?
> > >
> > > No, memset(&rte_eth_dev_data[port_id], ...) breaks it.
> >
> > Oh, you were talking about that particular case Remy's patch meant to
> > fix.
> >
> > > >> and fix is known.
> > > >
> > > > And there is already a fix?
> > >
> > > http://dpdk.org/ml/archives/dev/2017-January/054422.html
> >
> > Yes, it should fix that issue.
> 
> Well, few more thoughts: it may fix the crash issue Remy saw, but it
> looks like more a workaround to me. Basically, if primary and secondary
> shares a same port id, they should point to same device. Otherwise,
> primary process may use eth_dev->data for a device A, while the
> secondary process may use it for another device, as you said, it
> could be a vdev.
> 
> In such case, there is no way we could continue safely. That said,
> the given patch avoids the total reset of eth_dev->data, while it
> continues reset the eth_dev->data->name, which is wrong.
> 
> So it's not a proper fix.
> 
> Again, I think it's more about the usage. If primary starts with
> a nic device A, while the secondary starts with a nic device B,
> there is no way they could work well (unless they use different
> port id).

Why not?
I think this is possible.
They just need to be initialized properly,
so each rte_eth_devices[port_id]->data, etc. point to the right place.
Konstantin


> 
>   --yliu
> 
> > One question: do Remy or you regularly
> > run some multiple process test cases (and with vdev both in primary
> > and secondary process)?

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Yuanhan Liu]

On Mon, Jan 23, 2017 at 12:44:11PM +, Ananyev, Konstantin wrote:
> > On Mon, Jan 23, 2017 at 07:40:50PM +0800, Yuanhan Liu wrote:
> > > On Mon, Jan 23, 2017 at 11:32:23AM +, Ferruh Yigit wrote:
> > > > On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
> > > > > On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
> > > >   lib/librte_ether/rte_ethdev.c | 2 +-
> > > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > > 
> > > >  diff --git a/lib/librte_ether/rte_ethdev.c 
> > > >  b/lib/librte_ether/rte_ethdev.c
> > > >  index 4790faf..61f44e2 100644
> > > >  --- a/lib/librte_ether/rte_ethdev.c
> > > >  +++ b/lib/librte_ether/rte_ethdev.c
> > > >  @@ -225,7 +225,7 @@ struct rte_eth_dev *
> > > > return NULL;
> > > > }
> > > > 
> > > >  -  memset(&rte_eth_devices[port_id], 0, 
> > > >  sizeof(*eth_dev->data));
> > > >  +  memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
> > > >  rte_eth_dev_data));
> > > > >>>
> > > > >>> Not directly related to the this issue, but, after fix, this 
> > > > >>> may have
> > > > >>> issues with secondary process.
> > > > >>>
> > > > >>> There were patches sent to fix this.
> > > > >>
> > > > >> I mean this one:
> > > > >> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > > > >
> > > > > d948f596fee2 ("ethdev: fix port data mismatched in multiple 
> > > > > process
> > > > > model") should have fixed it.
> > > > 
> > > >  Think about case, where secondary process uses a virtual PMD, 
> > > >  which does
> > > >  a rte_eth_dev_allocate() call, shouldn't this corrupt primary 
> > > >  process
> > > >  device data?
> > > > >>>
> > > > >>> Yes, it may. However, I doubt that's the typical usage.
> > > > >>
> > > > >> But this is a use case, and broken now,
> > > > >
> > > > > I thought it was broken since the beginning?
> > > >
> > > > No, memset(&rte_eth_dev_data[port_id], ...) breaks it.
> > >
> > > Oh, you were talking about that particular case Remy's patch meant to
> > > fix.
> > >
> > > > >> and fix is known.
> > > > >
> > > > > And there is already a fix?
> > > >
> > > > http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > >
> > > Yes, it should fix that issue.
> > 
> > Well, few more thoughts: it may fix the crash issue Remy saw, but it
> > looks like more a workaround to me. Basically, if primary and secondary
> > shares a same port id, they should point to same device. Otherwise,
> > primary process may use eth_dev->data for a device A, while the
> > secondary process may use it for another device, as you said, it
> > could be a vdev.
> > 
> > In such case, there is no way we could continue safely. That said,
> > the given patch avoids the total reset of eth_dev->data, while it
> > continues reset the eth_dev->data->name, which is wrong.
> > 
> > So it's not a proper fix.
> > 
> > Again, I think it's more about the usage. If primary starts with
> > a nic device A, while the secondary starts with a nic device B,
> > there is no way they could work well (unless they use different
> > port id).
> 
> Why not?
> I think this is possible.

Yes, it's possible: find another port id if that one is already taken
by primary process (or even by secondary process: think that primary
process might attatch a port later).

> They just need to be initialized properly,
> so each rte_eth_devices[port_id]->data, etc. point to the right place.

My understanding is, as far as they use different port_id, it might
be fine. Just not sure it's enough or not.

--yliu

Re: [dpdk-dev] [PATCH v6 1/2] kni: add bulk function to free mbufs

[Olivier Matz]

Hi,

On Thu, 19 Jan 2017 11:46:58 +0700, Sergey Vyazmitinov
 wrote:
> Suggested-by: Stephen Hemminger 
> Signed-off-by: Sergey Vyazmitinov 
> ---
> v3:
> * Fixed issue with possible different mempools in buffer list.
> * Fixed issue with wrong rte_pktmbuf_alloc_bulk function return value
> processing in the kni_allocate_mbufs.
> ---
>  lib/librte_mbuf/rte_mbuf.h | 49
> ++ 1 file changed, 49
> insertions(+)
> 
> diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> index 4476d75..69d314f 100644
> --- a/lib/librte_mbuf/rte_mbuf.h
> +++ b/lib/librte_mbuf/rte_mbuf.h
> @@ -306,6 +306,9 @@ extern "C" {
>  /** Alignment constraint of mbuf private area. */
>  #define RTE_MBUF_PRIV_ALIGN 8
>  
> +/** Maximum number of mbufs freed in bulk. */
> +#define RTE_MBUF_BULK_FREE 64
> +
>  /**
>   * Get the name of a RX offload flag
>   *
> @@ -1261,6 +1264,52 @@ static inline void rte_pktmbuf_free(struct
> rte_mbuf *m) }
>  
>  /**
> + * Free n packets mbuf back into its original mempool.
> + *
> + * Free each mbuf, and all its segments in case of chained buffers.
> Each
> + * segment is added back into its original mempool.
> + *
> + * @param mp
> + *   The packets mempool.

This parameter was removed, it could be removed from the API comment.


> + * @param mbufs
> + *   The packets mbufs array to be freed.
> + * @param n
> + *   Number of packets.
> + */
> +static inline void rte_pktmbuf_free_bulk(struct rte_mbuf **mbufs,
> + unsigned int n)

I suggest we could use mbuf instead of pktmbuf in the function name.
It's a bit shorter, and the function would also apply on ctrlmbuf.

Also, the struct rte_mbuf **mbufs could probably be replaced by struct
rte_mbuf * const *mbufs.


> +{
> + void *tofree[RTE_MBUF_BULK_FREE];
> + struct rte_mempool *mp = NULL;
> + unsigned int i, count = 0;
> +
> + for (i = 0; i < n; i++) {
> + struct rte_mbuf *m, *m_next;
> +
> + for (m = mbufs[i]; m; m = m_next) {
> + m_next = m->next;
> +
> + if (count > 0 &&
> + (unlikely(m->pool != mp ||
> + count == RTE_MBUF_BULK_FREE))) {
> + rte_mempool_put_bulk(mp, tofree,
> count);
> + count = 0;
> + }
> +
> + mp = m->pool;
> +
> + if (likely(__rte_pktmbuf_prefree_seg(m) !=
> NULL)) {
> + m->next = NULL;
> + tofree[count++] = m;
> + }
> + }
> + }
> +
> + if (likely(count > 0))
> + rte_mempool_put_bulk(mp, tofree, count);
> +}
> +
> +/**
>   * Creates a "clone" of the given packet mbuf.
>   *

The function looks good to me, thank you. It looks also better than
what I've suggested in [1], since it properly manage mbuf chains. On
the other hand, I think my proposal could also help in drivers, where
segments are already unchained. I'll submit it in a RFC.

[1] http://dpdk.org/ml/archives/dev/2017-January/054538.html


One more thing, maybe it's worth adding a basic test in
app/test/test_mbuf.c.

Thanks,
Olivier

[dpdk-dev] [PATCH] doc: announce API/ABI changes for vhost

[Yuanhan Liu]

I made a vhost ABI/API refactoring at v16.04, meant to avoid such issue
forever. Well, apparently, I lied.

People are looking for more vhost-user options now days, other than
vhost-user net only. For example, SPDK (Storage Performance Development
Kit) are looking for chance of vhost-user SCSI and vhost-user block.

Apparently, they also need a vhost-user backend, while DPDK already
has a (mature enough) backend, they don't want to implement it again
from scratch. They want to leverage the one DPDK provides.

However, the last refactoring hasn't done that right, at least it's
not friendly for extending vhost-user to add more devices support.
For example, different virtio devices has its own feature set, while
APIs like rte_vhost_feature_disable(feature_mask) have no option to
tell the device type. Thus, a more proper API should look like:

rte_vhost_feature_disable(device_type, feature_mask);

Besides that, few public files and structures should be renamed, to
not let it bind to virtio-net. Specifically, they are:

- virtio_net_device_ops --> vhost_device_ops
- rte_virtio_net.h  --> rte_vhost.h

Signed-off-by: Yuanhan Liu 
---

I intended to send out the code first (I have already finished the
better part of it), while I'm starting vacation since tomorrow, leaving
me no time for that.
---
 doc/guides/rel_notes/deprecation.rst | 9 +
 1 file changed, 9 insertions(+)

diff --git a/doc/guides/rel_notes/deprecation.rst 
b/doc/guides/rel_notes/deprecation.rst
index 755dc65..5d6e9b6 100644
--- a/doc/guides/rel_notes/deprecation.rst
+++ b/doc/guides/rel_notes/deprecation.rst
@@ -62,3 +62,12 @@ Deprecation Notices
   PMDs that implement the latter.
   Target release for removal of the legacy API will be defined once most
   PMDs have switched to rte_flow.
+
+* vhost: API/ABI changes are planned for 17.05, for making DPDK vhost library
+  generic enough so that applications can build different vhost-user drivers
+  (instead of vhost-user net only) on top of that.
+  Specifically, ``virtio_net_device_ops`` will be renamed to 
``vhost_device_ops``.
+  Correspondingly, some API's parameter need be changed. Few more functions 
also
+  need be reworked to let it be device aware. For example, different virtio 
device
+  has different feature set, meaning functions like 
``rte_vhost_feature_disable``
+  need be changed. Last, file rte_virtio_net.h will be renamed to rte_vhost.h.
-- 
1.9.0

[dpdk-dev] [RFC] Add GRO support in DPDK

[Jiayu Hu]

With the support of hardware segmentation techniques in DPDK, the
networking stack overheads of send-side of applications, which directly
leverage DPDK, have been greatly reduced. But for receive-side, numbers of
segmented packets seriously burden the networking stack of applications.
Generic Receive Offload (GRO) is a widely used method to solve the
receive-side issue, which gains performance by reducing the amount of
packets processed by the networking stack. But currently, DPDK doesn't
support GRO. Therefore, we propose to add GRO support in DPDK, and this
RFC is used to explain the basic DPDK GRO design.

DPDK GRO is a SW-based packets assembly library, which provides GRO
abilities for numbers of protocols. In DPDK GRO, packets are merged
before returning to applications and after receiving from drivers.

In DPDK, GRO is a capability of NIC drivers. That support GRO or not and
what GRO types are supported are up to NIC drivers. Different drivers may
support different GRO types. By default, drivers enable all supported GRO
types. For applications, they can inquire the supported GRO types by
each driver, and can control what GRO types are applied. For example,
ixgbe supports TCP and UDP GRO, but the application just needs TCP GRO.
The application can disable ixgbe UDP GRO.

To support GRO, a driver should provide a way to tell applications what
GRO types are supported, and provides a GRO function, which is in charge
of assembling packets. Since different drivers may support different GRO
types, their GRO functions may be different. For applications, they don't
need extra operations to enable GRO. But if there are some GRO types that
are not needed, applications can use an API, like
rte_eth_gro_disable_protocols, to disable them. Besides, they can
re-enable the disabled ones.

The GRO function processes numbers of packets at a time. In each
invocation, what GRO types are applied depends on applications, and the
amount of packets to merge depends on the networking status and
applications. Specifically, applications determine the maximum number of
packets to be processed by the GRO function, but how many packets are
actually processed depends on if there are available packets to receive.
For example, the receive-side application asks the GRO function to
process 64 packets, but the sender only sends 40 packets. At this time,
the GRO function returns after processing 40 packets. To reassemble the
given packets, the GRO function performs an "assembly procedure" on each
packet. We use an example to demonstrate this procedure. Supposing the
GRO function is going to process packetX, it will do the following two
things:
a. Find a L4 assembly function according to the packet type of
packetX. A L4 assembly function is in charge of merging packets of a
specific type. For example, TCPv4 assembly function merges packets
whose L3 IPv4 and L4 is TCP. Each L4 assembly function has a packet
array, which keeps the packets that are unable to assemble.
Initially, the packet array is empty;
b. The L4 assembly function traverses own packet array to find a
mergeable packet (comparing Ethernet, IP and L4 header fields). If
finds, merges it and packetX via chaining them together; if doesn't,
allocates a new array element to store packetX and updates element
number of the array.
After performing the assembly procedure to all packets, the GRO function
combines the results of all packet arrays, and returns these packets to
applications.

There are lots of ways to implement the above design in DPDK. One of the
ways is:
a. Drivers tell applications what GRO types are supported via
dev->dev_ops->dev_infos_get;
b. When initialize, drivers register own GRO function as a RX
callback, which is invoked inside rte_eth_rx_burst. The name of the
GRO function should be like xxx_gro_receive (e.g. ixgbe_gro_receive).
Currently, the RX callback can only process the packets returned by
dev->rx_pkt_burst each time, and the maximum packet number
dev->rx_pkt_burst returns is determined by each driver, which can't
be interfered by applications. Therefore, to implement the above GRO
design, we have to modify current RX implementation to make driver
return packets as many as possible until the packet number meets the
demand of applications or there are not available packets to receive.
This modification is also proposed in patch:
http://dpdk.org/ml/archives/dev/2017-January/055887.html;
c. The GRO types to apply and the maximum number of packets to merge
are passed by resetting RX callback parameters. It can be achieved by
invoking rte_eth_rx_callback;
d. Simply, we can just store packet addresses into the packet array.
To check one element, we need to fetch the packet via its address.
However, this simple desig

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Ananyev, Konstantin]

> -Original Message-
> From: Yuanhan Liu [mailto:yuanhan@linux.intel.com]
> Sent: Monday, January 23, 2017 12:53 PM
> To: Ananyev, Konstantin 
> Cc: Yigit, Ferruh ; dev@dpdk.org; Thomas Monjalon 
> ; Horton, Remy
> 
> Subject: Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset
> 
> On Mon, Jan 23, 2017 at 12:44:11PM +, Ananyev, Konstantin wrote:
> > > On Mon, Jan 23, 2017 at 07:40:50PM +0800, Yuanhan Liu wrote:
> > > > On Mon, Jan 23, 2017 at 11:32:23AM +, Ferruh Yigit wrote:
> > > > > On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
> > > > > > On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
> > > > >   lib/librte_ether/rte_ethdev.c | 2 +-
> > > > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > 
> > > > >  diff --git a/lib/librte_ether/rte_ethdev.c 
> > > > >  b/lib/librte_ether/rte_ethdev.c
> > > > >  index 4790faf..61f44e2 100644
> > > > >  --- a/lib/librte_ether/rte_ethdev.c
> > > > >  +++ b/lib/librte_ether/rte_ethdev.c
> > > > >  @@ -225,7 +225,7 @@ struct rte_eth_dev *
> > > > >   return NULL;
> > > > >   }
> > > > > 
> > > > >  -memset(&rte_eth_devices[port_id], 0, 
> > > > >  sizeof(*eth_dev->data));
> > > > >  +memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
> > > > >  rte_eth_dev_data));
> > > > > >>>
> > > > > >>> Not directly related to the this issue, but, after fix, this 
> > > > > >>> may have
> > > > > >>> issues with secondary process.
> > > > > >>>
> > > > > >>> There were patches sent to fix this.
> > > > > >>
> > > > > >> I mean this one:
> > > > > >> http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > > > > >
> > > > > > d948f596fee2 ("ethdev: fix port data mismatched in multiple 
> > > > > > process
> > > > > > model") should have fixed it.
> > > > > 
> > > > >  Think about case, where secondary process uses a virtual PMD, 
> > > > >  which does
> > > > >  a rte_eth_dev_allocate() call, shouldn't this corrupt primary 
> > > > >  process
> > > > >  device data?
> > > > > >>>
> > > > > >>> Yes, it may. However, I doubt that's the typical usage.
> > > > > >>
> > > > > >> But this is a use case, and broken now,
> > > > > >
> > > > > > I thought it was broken since the beginning?
> > > > >
> > > > > No, memset(&rte_eth_dev_data[port_id], ...) breaks it.
> > > >
> > > > Oh, you were talking about that particular case Remy's patch meant to
> > > > fix.
> > > >
> > > > > >> and fix is known.
> > > > > >
> > > > > > And there is already a fix?
> > > > >
> > > > > http://dpdk.org/ml/archives/dev/2017-January/054422.html
> > > >
> > > > Yes, it should fix that issue.
> > >
> > > Well, few more thoughts: it may fix the crash issue Remy saw, but it
> > > looks like more a workaround to me. Basically, if primary and secondary
> > > shares a same port id, they should point to same device. Otherwise,
> > > primary process may use eth_dev->data for a device A, while the
> > > secondary process may use it for another device, as you said, it
> > > could be a vdev.
> > >
> > > In such case, there is no way we could continue safely. That said,
> > > the given patch avoids the total reset of eth_dev->data, while it
> > > continues reset the eth_dev->data->name, which is wrong.
> > >
> > > So it's not a proper fix.
> > >
> > > Again, I think it's more about the usage. If primary starts with
> > > a nic device A, while the secondary starts with a nic device B,
> > > there is no way they could work well (unless they use different
> > > port id).
> >
> > Why not?
> > I think this is possible.
> 
> Yes, it's possible: find another port id if that one is already taken
> by primary process (or even by secondary process: think that primary
> process might attatch a port later).
> 
> > They just need to be initialized properly,
> > so each rte_eth_devices[port_id]->data, etc. point to the right place.
> 
> My understanding is, as far as they use different port_id, it might
> be fine. Just not sure it's enough or not.

As I understand, the main problem is that  rte_eth_devices[] is local,
while rte_eth_dev_data points to the shared memory array.
And rte_eth_dev_allocate() assumes that if rte_eth_devices[x] is free,
then rte_eth_dev_data[port_id] is also free.
Which is wrong in case when primary/secondary processes have different devices 
attached.
Another problem is that inside rte_ethdev.c we manipulate rte_eth_dev_data[]
contents without grabbing any lock.
I think it was an attempt to fix that issue in 16.07 timeframe or so,
but I don't remember what happened with that patch.
Konstantin 

Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset

[Ferruh Yigit]

On 1/23/2017 1:06 PM, Ananyev, Konstantin wrote:
> 
> 
>> -Original Message-
>> From: Yuanhan Liu [mailto:yuanhan@linux.intel.com]
>> Sent: Monday, January 23, 2017 12:53 PM
>> To: Ananyev, Konstantin 
>> Cc: Yigit, Ferruh ; dev@dpdk.org; Thomas Monjalon 
>> ; Horton, Remy
>> 
>> Subject: Re: [dpdk-dev] [PATCH] ethdev: fix wrong memset
>>
>> On Mon, Jan 23, 2017 at 12:44:11PM +, Ananyev, Konstantin wrote:
 On Mon, Jan 23, 2017 at 07:40:50PM +0800, Yuanhan Liu wrote:
> On Mon, Jan 23, 2017 at 11:32:23AM +, Ferruh Yigit wrote:
>> On 1/23/2017 11:24 AM, Yuanhan Liu wrote:
>>> On Mon, Jan 23, 2017 at 11:05:25AM +, Ferruh Yigit wrote:
>>  lib/librte_ether/rte_ethdev.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/lib/librte_ether/rte_ethdev.c 
>> b/lib/librte_ether/rte_ethdev.c
>> index 4790faf..61f44e2 100644
>> --- a/lib/librte_ether/rte_ethdev.c
>> +++ b/lib/librte_ether/rte_ethdev.c
>> @@ -225,7 +225,7 @@ struct rte_eth_dev *
>>  return NULL;
>>  }
>>
>> -memset(&rte_eth_devices[port_id], 0, 
>> sizeof(*eth_dev->data));
>> +memset(&rte_eth_dev_data[port_id], 0, sizeof(struct 
>> rte_eth_dev_data));
>
> Not directly related to the this issue, but, after fix, this may 
> have
> issues with secondary process.
>
> There were patches sent to fix this.

 I mean this one:
 http://dpdk.org/ml/archives/dev/2017-January/054422.html
>>>
>>> d948f596fee2 ("ethdev: fix port data mismatched in multiple process
>>> model") should have fixed it.
>>
>> Think about case, where secondary process uses a virtual PMD, which 
>> does
>> a rte_eth_dev_allocate() call, shouldn't this corrupt primary process
>> device data?
>
> Yes, it may. However, I doubt that's the typical usage.

 But this is a use case, and broken now,
>>>
>>> I thought it was broken since the beginning?
>>
>> No, memset(&rte_eth_dev_data[port_id], ...) breaks it.
>
> Oh, you were talking about that particular case Remy's patch meant to
> fix.
>
 and fix is known.
>>>
>>> And there is already a fix?
>>
>> http://dpdk.org/ml/archives/dev/2017-January/054422.html
>
> Yes, it should fix that issue.

 Well, few more thoughts: it may fix the crash issue Remy saw, but it
 looks like more a workaround to me. Basically, if primary and secondary
 shares a same port id, they should point to same device. Otherwise,
 primary process may use eth_dev->data for a device A, while the
 secondary process may use it for another device, as you said, it
 could be a vdev.

 In such case, there is no way we could continue safely. That said,
 the given patch avoids the total reset of eth_dev->data, while it
 continues reset the eth_dev->data->name, which is wrong.

 So it's not a proper fix.

 Again, I think it's more about the usage. If primary starts with
 a nic device A, while the secondary starts with a nic device B,
 there is no way they could work well (unless they use different
 port id).
>>>
>>> Why not?
>>> I think this is possible.
>>
>> Yes, it's possible: find another port id if that one is already taken
>> by primary process (or even by secondary process: think that primary
>> process might attatch a port later).
>>
>>> They just need to be initialized properly,
>>> so each rte_eth_devices[port_id]->data, etc. point to the right place.
>>
>> My understanding is, as far as they use different port_id, it might
>> be fine. Just not sure it's enough or not.
> 
> As I understand, the main problem is that  rte_eth_devices[] is local,
> while rte_eth_dev_data points to the shared memory array.
> And rte_eth_dev_allocate() assumes that if rte_eth_devices[x] is free,
> then rte_eth_dev_data[port_id] is also free.
> Which is wrong in case when primary/secondary processes have different 
> devices attached.
> Another problem is that inside rte_ethdev.c we manipulate rte_eth_dev_data[]
> contents without grabbing any lock.

> I think it was an attempt to fix that issue in 16.07 timeframe or so,
> but I don't remember what happened with that patch.

Same here, I remember this already discussed and even some patches sent,
by Reshma if I remember correctly, but I don't remember latest status.

> Konstantin 
> 
> 
> 
> 

Re: [dpdk-dev] [PATCH v6 1/2] kni: add bulk function to free mbufs

[Olivier Matz]

Hi,

[resent and updated, previous mail was sent from the wrong address]


On Mon, 23 Jan 2017 13:59:47 +0100, Olivier Matz
 wrote:
> > + * @param mbufs
> > + *   The packets mbufs array to be freed.
> > + * @param n
> > + *   Number of packets.
> > + */
> > +static inline void rte_pktmbuf_free_bulk(struct rte_mbuf **mbufs,
> > +   unsigned int n)  
> 
> I suggest we could use mbuf instead of pktmbuf in the function name.
> It's a bit shorter, and the function would also apply on ctrlmbuf.

Sorry, please ignore this one. It's more consistent to have a
rte_pktmbuf_free and rte_pktmbuf_free_bulk having the same prefix.



Also a small comment on the title: it should be "mbuf:", not "kni:".

In the other patch, the uppercase in "Use" could also be changed to
lowercase (seen with ./devtools/check-git-log.sh).

Thanks,
Olivier

Re: [dpdk-dev] [PATCH] net/qede/base: fix freebsd compilation

[Ferruh Yigit]

On 1/23/2017 10:21 AM, Thomas Monjalon wrote:
> 2017-01-22 21:02, Rasesh Mody:
>> This patch addresses compilation errors on FreeBSD 11.0 clang 3.8.0.
> 
> Is it really specific to FreeBSD 11? I guess it is more about clang.
> 

Right, this seem clang 3.8.0 issue, independent from FreeBSD version.

I will remove FreeBSD version information while applying.

Re: [dpdk-dev] [PATCH] net/qede/base: fix freebsd compilation

[Ferruh Yigit]

On 1/23/2017 5:02 AM, Rasesh Mody wrote:
> This patch addresses compilation errors on FreeBSD 11.0 clang 3.8.0.
> 
> Fixes: ec94dbc57362 ("qede: add base driver")
> 
> Signed-off-by: Rasesh Mody 

Applied to dpdk-next-net/master, thanks.

(FreeBSD version information removed from commit)

Re: [dpdk-dev] [PATCH v3 2/3] net/e1000: e1000 igb support to free consumed buffers

[Billy McFall]

There are two scenarios the APP calls this API (there may be others, but
the API was designed with these two in mind):

1) APP receives a packet that needs to be flooded. Instead of making a copy
of the packet (whether it is the whole packet, just the header or a clone
buffer is used, as done today), the APP increments the reference count,
adjusts the header and then sends the packet to one of the destination
ports. The APP then polls this API until the reference count is
decremented, and forwards to the next destination port with any header
modifications needed.

2) An APP like a packet generation completes a run and then wants to reset
for another run. It can call this API to free and mbufs that have been
cache by the driver so it can start another run with the full set of mbufs.
Maybe it is switching interfaces that are on another core and want to
reclaim any mbufs that have been cached.

So this API is intended to be called on an as need basis, not on every
packet. When either a particular mbuf is desired or the mbufs for a
particular interface need to be released.

Hope this helps,
Billy McFall

On Sat, Jan 21, 2017 at 10:47 PM, Lu, Wenzhuo  wrote:

> Hi Billy,
>
> > -Original Message-
> > From: Billy McFall [mailto:bmcf...@redhat.com]
> > Sent: Saturday, January 21, 2017 12:01 AM
> > To: thomas.monja...@6wind.com; Lu, Wenzhuo
> > Cc: dev@dpdk.org; Billy McFall
> > Subject: [PATCH v3 2/3] net/e1000: e1000 igb support to free consumed
> > buffers
> >
> > Add support to the e1000 igb driver for the new API to force free
> consumed
> > buffers on Tx ring. e1000 igb driver does not implement a tx_rs_thresh
> to free
> > mbufs, it frees a slot in the ring as needed, so a new function needed
> to be
> > written.
> >
> > Signed-off-by: Billy McFall 
> > ---
> >  drivers/net/e1000/e1000_ethdev.h |   2 +
> >  drivers/net/e1000/igb_ethdev.c   |   1 +
> >  drivers/net/e1000/igb_rxtx.c | 126
> > +++
> >  3 files changed, 129 insertions(+)
> >
> > diff --git a/drivers/net/e1000/e1000_ethdev.h
> > b/drivers/net/e1000/e1000_ethdev.h
> > index 81a6dbb..39b2f43 100644
> > --- a/drivers/net/e1000/e1000_ethdev.h
> > +++ b/drivers/net/e1000/e1000_ethdev.h
> > @@ -315,6 +315,8 @@ int eth_igb_tx_queue_setup(struct rte_eth_dev *dev,
> > uint16_t tx_queue_id,
> >   uint16_t nb_tx_desc, unsigned int socket_id,
> >   const struct rte_eth_txconf *tx_conf);
> >
> > +int eth_igb_tx_done_cleanup(void *txq, uint32_t free_cnt);
> > +
> >  int eth_igb_rx_init(struct rte_eth_dev *dev);
> >
> >  void eth_igb_tx_init(struct rte_eth_dev *dev); diff --git
> > a/drivers/net/e1000/igb_ethdev.c b/drivers/net/e1000/igb_ethdev.c index
> > 8843dd1..12a1a30 100644
> > --- a/drivers/net/e1000/igb_ethdev.c
> > +++ b/drivers/net/e1000/igb_ethdev.c
> > @@ -408,6 +408,7 @@ static const struct eth_dev_ops eth_igb_ops = {
> >   .rx_descriptor_done   = eth_igb_rx_descriptor_done,
> >   .tx_queue_setup   = eth_igb_tx_queue_setup,
> >   .tx_queue_release = eth_igb_tx_queue_release,
> > + .tx_done_cleanup  = eth_igb_tx_done_cleanup,
> >   .dev_led_on   = eth_igb_led_on,
> >   .dev_led_off  = eth_igb_led_off,
> >   .flow_ctrl_get= eth_igb_flow_ctrl_get,
> > diff --git a/drivers/net/e1000/igb_rxtx.c b/drivers/net/e1000/igb_rxtx.c
> index
> > 45f3f24..00cd2aa 100644
> > --- a/drivers/net/e1000/igb_rxtx.c
> > +++ b/drivers/net/e1000/igb_rxtx.c
> > @@ -1277,6 +1277,132 @@ eth_igb_tx_queue_release(void *txq)
> >   igb_tx_queue_release(txq);
> >  }
> >
> > +static int
> > +igb_tx_done_cleanup(struct igb_tx_queue *txq, uint32_t free_cnt) {
> > + struct igb_tx_entry *sw_ring;
> > + volatile union e1000_adv_tx_desc *txr;
> > + uint16_t tx_first; /* First segment analyzed. */
> > + uint16_t tx_id;/* Current segment being processed. */
> > + uint16_t tx_last;  /* Last segment in the current packet. */
> > + uint16_t tx_next;  /* First segment of the next packet. */
> > + int count;
> > +
> > + if (txq != NULL) {
> > + count = 0;
> > + sw_ring = txq->sw_ring;
> > + txr = txq->tx_ring;
> > +
> > + /*
> > +  * tx_tail is the last sent packet on the sw_ring. Goto
> the end
> > +  * of that packet (the last segment in the packet chain)
> and
> > +  * then the next segment will be the start of the oldest
> > segment
> > +  * in the sw_ring. This is the first packet that will be
> > +  * attempted to be freed.
> > +  */
> > +
> > + /* Get last segment in most recently added packet. */
> > + tx_first = sw_ring[txq->tx_tail].last_id;
> > +
> > + /* Get the next segment, which is the oldest segment in
> ring.
> > */
> > + tx_first = sw_ring[tx_first].next_id;
> > +
> > + /* Set the current index to the first. */
> > +   

Re: [dpdk-dev] [PATCH 1/2] net/i40e: fix MAC check when delete MAC

[Ferruh Yigit]

On 1/23/2017 9:42 AM, Jingjing Wu wrote:
> When VF sends request to remove MAC address, PF host will check
> if it is a non-zero or unicast address. When VF remove a multicast
> address, it will report error.
> This patch fixes this issue.
> 
> Fixes: ec852c94af39 ("net/i40e: enhance sanity check of MAC")
Cc: sta...@dpdk.org

> Signed-off-by: Jingjing Wu 

Series applied to dpdk-next-net/master, thanks.

[dpdk-dev] [PATCH] maintainers: update mlx4/mlx5

[Adrien Mazarguil]

Promote Nelio as additional maintainer for mlx4 and mlx5 PMDs.

Signed-off-by: Adrien Mazarguil 
Acked-by: Nelio Laranjeiro 
Acked-by: Olga Shern 
---
 MAINTAINERS | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index f071138..541c888 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -341,11 +341,13 @@ F: drivers/net/fm10k/
 
 Mellanox mlx4
 M: Adrien Mazarguil 
+M: Nelio Laranjeiro 
 F: drivers/net/mlx4/
 F: doc/guides/nics/mlx4.rst
 
 Mellanox mlx5
 M: Adrien Mazarguil 
+M: Nelio Laranjeiro 
 F: drivers/net/mlx5/
 F: doc/guides/nics/mlx5.rst
 
-- 
2.1.4

Re: [dpdk-dev] [PATCH v3 0/3] new API to free consumed buffers in Tx ring

[Thomas Monjalon]

Hi,

Thanks for bringing a new convenient API.

2017-01-20 11:01, Billy McFall:
> Based on a request from Damjan Marion and seconded by Keith Wiles, see
> dpdk-dev mailing list from 11/21/2016,

Generally we use links to ML archives to show prior discussions.

> add a new API to free consumed
> buffers on TX ring. This addresses two scenarios:
> 1) Flooding a packet and want to reuse existing mbuf to avoid a packet
> copy. Increment the reference count of the packet and poll new API until
> reference count is decremented.
> 2) Application runs out of mbufs, or resets and is preparing for
> additional run, call API to free consumed packets so processing can
> continue.
> 
> API will return the number of packets freed (0-n) or error code if
> feature not supported (-ENOTSUP) or input invalid (-ENODEV).
> 
> API for e1000 igb driver and vHost driver have been implemented. Other
> drivers can be implemented over time. Some drivers implement a Tx done
> flush routine that should be reused where possible. e1000 igb driver
> and vHost driver do not have such functions.

Before considering to use such API, user apps may check how well it
is supported by drivers in this table:
http://dpdk.org/doc/guides/nics/overview.html

Please add this feature in the appropriate files:
doc/guides/nics/features/default.ini in patch 1
doc/guides/nics/features/e1000.ini in patch 2
doc/guides/nics/features/vhost.ini in patch 3

Then it will be introduced in the beginning of the 17.05 cycle.
So more drivers could be implemented before the 17.05 release.
Does it sound good?

PS: please use --in-reply-to when sending new versions to keep track
of versions changes and discussions.

[dpdk-dev] transmit lockup on i210

[Dennis Montgomery]

Hello DPDK community,

I haven’t found any relevant errata/fix notes so I’m asking here. I’m running 
into a transmit lockup on an i210 NIC: the transmit descriptor ring fills up 
and never drains. I’ve instrumented my code to dump out all extended stats and 
registers and found that when the transmitter gets ‘stuck’, the E1000_TDH 
register pegs to a particular index, and TDT catches up to it.

I have the transmitter configured with these thresholds: pthresh == 8 hthresh 
== 1 wthresh == 16, four queues, 512 tx descriptors. I’m running DPDK 16.07.

When this happens, restarting my application (including a NIC unbind/bind 
cycle) fixes it up, so the hardware isn’t wedged.

I wonder if (a) anyone has run into a tx lockup with the IGB driver, and/or (b) 
what should I be looking at?

Regards, and thanks for any help,

Dennis Montgomery

[dpdk-dev] [PATCH v6 1/6] lib: distributor performance enhancements

[David Hunt]

Now sends bursts of up to 8 mbufs to each worker, and tracks
the in-flight flow-ids (atomic scheduling)

New file with a new api, similar to the old API except with _burst
at the end of the function names. This is to preserve the original
API (and code) for backward compatibility.

It uses a similar handshake mechanism to the previous version of
the library, in that bits are used to indicate when packets are ready
to be sent to a worker and ready to be returned from a worker. One main
difference is that instead of sending one packet in a cache line, it
makes use of the 7 free spaces in the same cache line in order to send
up to 8 packets at a time to/from a worker.

The flow matching algorithm has had significant re-work, and now keeps
an array of inflight flows and an array of backlog flows, and matches
incoming flows to the inflight/backlog flows of all workers so
that flow pinning to workers can be maintained.

Signed-off-by: David Hunt 
---
 lib/librte_distributor/Makefile|   2 +
 lib/librte_distributor/rte_distributor.c   |  74 +--
 lib/librte_distributor/rte_distributor.h   |   2 +-
 lib/librte_distributor/rte_distributor_burst.c | 564 +
 lib/librte_distributor/rte_distributor_burst.h | 257 ++
 lib/librte_distributor/rte_distributor_private.h   | 189 +++
 lib/librte_distributor/rte_distributor_version.map |  14 +
 7 files changed, 1029 insertions(+), 73 deletions(-)
 create mode 100644 lib/librte_distributor/rte_distributor_burst.c
 create mode 100644 lib/librte_distributor/rte_distributor_burst.h
 create mode 100644 lib/librte_distributor/rte_distributor_private.h

diff --git a/lib/librte_distributor/Makefile b/lib/librte_distributor/Makefile
index 4c9af17..2acc54d 100644
--- a/lib/librte_distributor/Makefile
+++ b/lib/librte_distributor/Makefile
@@ -43,9 +43,11 @@ LIBABIVER := 1
 
 # all source are stored in SRCS-y
 SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) := rte_distributor.c
+SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor_burst.c
 
 # install this header file
 SYMLINK-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR)-include := rte_distributor.h
+SYMLINK-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR)-include += rte_distributor_burst.h
 
 # this lib needs eal
 DEPDIRS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += lib/librte_eal
diff --git a/lib/librte_distributor/rte_distributor.c 
b/lib/librte_distributor/rte_distributor.c
index f3f778c..1bcee4c 100644
--- a/lib/librte_distributor/rte_distributor.c
+++ b/lib/librte_distributor/rte_distributor.c
@@ -1,7 +1,7 @@
 /*-
  *   BSD LICENSE
  *
- *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+ *   Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
  *   All rights reserved.
  *
  *   Redistribution and use in source and binary forms, with or without
@@ -40,79 +40,9 @@
 #include 
 #include 
 #include 
+#include "rte_distributor_private.h"
 #include "rte_distributor.h"
 
-#define NO_FLAGS 0
-#define RTE_DISTRIB_PREFIX "DT_"
-
-/* we will use the bottom four bits of pointer for flags, shifting out
- * the top four bits to make room (since a 64-bit pointer actually only uses
- * 48 bits). An arithmetic-right-shift will then appropriately restore the
- * original pointer value with proper sign extension into the top bits. */
-#define RTE_DISTRIB_FLAG_BITS 4
-#define RTE_DISTRIB_FLAGS_MASK (0x0F)
-#define RTE_DISTRIB_NO_BUF 0   /**< empty flags: no buffer requested */
-#define RTE_DISTRIB_GET_BUF (1)/**< worker requests a buffer, returns old 
*/
-#define RTE_DISTRIB_RETURN_BUF (2) /**< worker returns a buffer, no request */
-
-#define RTE_DISTRIB_BACKLOG_SIZE 8
-#define RTE_DISTRIB_BACKLOG_MASK (RTE_DISTRIB_BACKLOG_SIZE - 1)
-
-#define RTE_DISTRIB_MAX_RETURNS 128
-#define RTE_DISTRIB_RETURNS_MASK (RTE_DISTRIB_MAX_RETURNS - 1)
-
-/**
- * Maximum number of workers allowed.
- * Be aware of increasing the limit, becaus it is limited by how we track
- * in-flight tags. See @in_flight_bitmask and @rte_distributor_process
- */
-#define RTE_DISTRIB_MAX_WORKERS64
-
-/**
- * Buffer structure used to pass the pointer data between cores. This is cache
- * line aligned, but to improve performance and prevent adjacent cache-line
- * prefetches of buffers for other workers, e.g. when worker 1's buffer is on
- * the next cache line to worker 0, we pad this out to three cache lines.
- * Only 64-bits of the memory is actually used though.
- */
-union rte_distributor_buffer {
-   volatile int64_t bufptr64;
-   char pad[RTE_CACHE_LINE_SIZE*3];
-} __rte_cache_aligned;
-
-struct rte_distributor_backlog {
-   unsigned start;
-   unsigned count;
-   int64_t pkts[RTE_DISTRIB_BACKLOG_SIZE];
-};
-
-struct rte_distributor_returned_pkts {
-   unsigned start;
-   unsigned count;
-   struct rte_mbuf *mbufs[RTE_DISTRIB_MAX_RETURNS];
-};
-
-struct rte_distributor {
-   TAILQ_ENTRY(rte_distributor) next;/**< Next in list. */
-
-   char name[RTE_DISTRIBUTOR_NAME

[dpdk-dev] [PATCH v6 2/6] lib: add distributor vector flow matching

[David Hunt]

Signed-off-by: David Hunt 
---
 lib/librte_distributor/Makefile|   7 ++
 lib/librte_distributor/rte_distributor_burst.c |  12 ++-
 .../rte_distributor_match_generic.c|  43 
 lib/librte_distributor/rte_distributor_match_sse.c | 113 +
 lib/librte_distributor/rte_distributor_private.h   |  13 +++
 5 files changed, 186 insertions(+), 2 deletions(-)
 create mode 100644 lib/librte_distributor/rte_distributor_match_generic.c
 create mode 100644 lib/librte_distributor/rte_distributor_match_sse.c

diff --git a/lib/librte_distributor/Makefile b/lib/librte_distributor/Makefile
index 2acc54d..4baaa0c 100644
--- a/lib/librte_distributor/Makefile
+++ b/lib/librte_distributor/Makefile
@@ -44,6 +44,13 @@ LIBABIVER := 1
 # all source are stored in SRCS-y
 SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) := rte_distributor.c
 SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor_burst.c
+ifeq ($(CONFIG_RTE_ARCH_X86),y)
+SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor_match_sse.c
+CFLAGS_rte_distributor_match_sse.o += -msse4.2
+else
+SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor_match_generic.c
+endif
+
 
 # install this header file
 SYMLINK-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR)-include := rte_distributor.h
diff --git a/lib/librte_distributor/rte_distributor_burst.c 
b/lib/librte_distributor/rte_distributor_burst.c
index 9315c12..c000b8d 100644
--- a/lib/librte_distributor/rte_distributor_burst.c
+++ b/lib/librte_distributor/rte_distributor_burst.c
@@ -190,7 +190,7 @@ store_return(uintptr_t oldbuf, struct rte_distributor_burst 
*d,
  * backlog). This will then allow us to pin those packets to the relevant
  * workers to give us our atomic flow pinning.
  */
-static inline void
+void
 find_match_scalar(struct rte_distributor_burst *d,
uint16_t *data_ptr,
uint16_t *output_ptr)
@@ -357,6 +357,9 @@ rte_distributor_process_burst(struct rte_distributor_burst 
*d,
flows[i] = 0;
 
switch (d->dist_match_fn) {
+   case RTE_DIST_MATCH_VECTOR:
+   find_match_vec(d, &flows[0], &matches[0]);
+   break;
default:
find_match_scalar(d, &flows[0], &matches[0]);
}
@@ -544,7 +547,12 @@ rte_distributor_create_burst(const char *name,
snprintf(d->name, sizeof(d->name), "%s", name);
d->num_workers = num_workers;
 
-   d->dist_match_fn = RTE_DIST_MATCH_SCALAR;
+#if defined(RTE_ARCH_X86)
+   if (rte_cpu_get_flag_enabled(RTE_CPUFLAG_SSE4_2)) {
+   d->dist_match_fn = RTE_DIST_MATCH_VECTOR;
+   } else
+#endif
+   d->dist_match_fn = RTE_DIST_MATCH_SCALAR;
 
/*
 * Set up the backog tags so they're pointing at the second cache
diff --git a/lib/librte_distributor/rte_distributor_match_generic.c 
b/lib/librte_distributor/rte_distributor_match_generic.c
new file mode 100644
index 000..a523f08
--- /dev/null
+++ b/lib/librte_distributor/rte_distributor_match_generic.c
@@ -0,0 +1,43 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2017 Intel Corporation. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include 
+#include "rte_distributor_private.h"
+#include "rte_distributor_burst.h"
+
+void
+find_match_vec(struct rte_distributor_burst *d,
+   uint16_t *data_p

[dpdk-dev] [PATCH v6 4/6] test: add distributor perf autotest

[David Hunt]

Signed-off-by: David Hunt 
---
 app/test/test_distributor_perf.c | 150 +++
 1 file changed, 138 insertions(+), 12 deletions(-)

diff --git a/app/test/test_distributor_perf.c b/app/test/test_distributor_perf.c
index 7947fe9..9132010 100644
--- a/app/test/test_distributor_perf.c
+++ b/app/test/test_distributor_perf.c
@@ -1,7 +1,7 @@
 /*-
  *   BSD LICENSE
  *
- *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+ *   Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
  *   All rights reserved.
  *
  *   Redistribution and use in source and binary forms, with or without
@@ -40,9 +40,11 @@
 #include 
 #include 
 #include 
+#include 
 
-#define ITER_POWER 20 /* log 2 of how many iterations we do when timing. */
-#define BURST 32
+#define ITER_POWER_CL 25 /* log 2 of how many iterations  for Cache Line test 
*/
+#define ITER_POWER 21 /* log 2 of how many iterations we do when timing. */
+#define BURST 64
 #define BIG_BATCH 1024
 
 /* static vars - zero initialized by default */
@@ -54,7 +56,8 @@ struct worker_stats {
 } __rte_cache_aligned;
 struct worker_stats worker_stats[RTE_MAX_LCORE];
 
-/* worker thread used for testing the time to do a round-trip of a cache
+/*
+ * worker thread used for testing the time to do a round-trip of a cache
  * line between two cores and back again
  */
 static void
@@ -69,7 +72,8 @@ flip_bit(volatile uint64_t *arg)
}
 }
 
-/* test case to time the number of cycles to round-trip a cache line between
+/*
+ * test case to time the number of cycles to round-trip a cache line between
  * two cores and back again.
  */
 static void
@@ -86,7 +90,7 @@ time_cache_line_switch(void)
rte_pause();
 
const uint64_t start_time = rte_rdtsc();
-   for (i = 0; i < (1 << ITER_POWER); i++) {
+   for (i = 0; i < (1 << ITER_POWER_CL); i++) {
while (*pdata)
rte_pause();
*pdata = 1;
@@ -98,13 +102,14 @@ time_cache_line_switch(void)
*pdata = 2;
rte_eal_wait_lcore(slaveid);
printf(" Cache line switch test ===\n");
-   printf("Time for %u iterations = %"PRIu64" ticks\n", (1<> ITER_POWER);
+   (end_time-start_time) >> ITER_POWER_CL);
 }
 
-/* returns the total count of the number of packets handled by the worker
+/*
+ * returns the total count of the number of packets handled by the worker
  * functions given below.
  */
 static unsigned
@@ -123,7 +128,8 @@ clear_packet_count(void)
memset(&worker_stats, 0, sizeof(worker_stats));
 }
 
-/* this is the basic worker function for performance tests.
+/*
+ * this is the basic worker function for performance tests.
  * it does nothing but return packets and count them.
  */
 static int
@@ -144,7 +150,37 @@ handle_work(void *arg)
return 0;
 }
 
-/* this basic performance test just repeatedly sends in 32 packets at a time
+/*
+ * this is the basic worker function for performance tests.
+ * it does nothing but return packets and count them.
+ */
+static int
+handle_work_burst(void *arg)
+{
+   struct rte_distributor_burst *d = arg;
+   unsigned int count = 0;
+   unsigned int num = 0;
+   int i;
+   unsigned int id = __sync_fetch_and_add(&worker_idx, 1);
+   struct rte_mbuf *buf[8] __rte_cache_aligned;
+
+   for (i = 0; i < 8; i++)
+   buf[i] = NULL;
+
+   num = rte_distributor_get_pkt_burst(d, id, buf, buf, num);
+   while (!quit) {
+   worker_stats[id].handled_packets += num;
+   count += num;
+   num = rte_distributor_get_pkt_burst(d, id, buf, buf, num);
+   }
+   worker_stats[id].handled_packets += num;
+   count += num;
+   rte_distributor_return_pkt_burst(d, id, buf, num);
+   return 0;
+}
+
+/*
+ * this basic performance test just repeatedly sends in 32 packets at a time
  * to the distributor and verifies at the end that we got them all in the 
worker
  * threads and finally how long per packet the processing took.
  */
@@ -174,6 +210,8 @@ perf_test(struct rte_distributor *d, struct rte_mempool *p)
rte_distributor_process(d, NULL, 0);
} while (total_packet_count() < (BURST << ITER_POWER));
 
+   rte_distributor_clear_returns(d);
+
printf("=== Performance test of distributor ===\n");
printf("Time per burst:  %"PRIu64"\n", (end - start) >> ITER_POWER);
printf("Time per packet: %"PRIu64"\n\n",
@@ -190,6 +228,55 @@ perf_test(struct rte_distributor *d, struct rte_mempool *p)
return 0;
 }
 
+/*
+ * this basic performance test just repeatedly sends in 32 packets at a time
+ * to the distributor and verifies at the end that we got them all in the 
worker
+ * threads and finally how long per packet the processing took.
+ */
+static inline int
+perf_test_burst(struct rte_distributor_burst *d, struct rte_mempool *p)
+{
+   unsigned int i;
+   uint64_t start, end;
+   struc

[dpdk-dev] [PATCH v6 3/6] test: unit tests for new distributor burst API

[David Hunt]

Signed-off-by: David Hunt 
---
 app/test/test_distributor.c | 501 ++--
 1 file changed, 392 insertions(+), 109 deletions(-)

diff --git a/app/test/test_distributor.c b/app/test/test_distributor.c
index 85cb8f3..3871f86 100644
--- a/app/test/test_distributor.c
+++ b/app/test/test_distributor.c
@@ -1,7 +1,7 @@
 /*-
  *   BSD LICENSE
  *
- *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+ *   Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
  *   All rights reserved.
  *
  *   Redistribution and use in source and binary forms, with or without
@@ -40,11 +40,24 @@
 #include 
 #include 
 #include 
+#include 
 
 #define ITER_POWER 20 /* log 2 of how many iterations we do when timing. */
 #define BURST 32
 #define BIG_BATCH 1024
 
+#define DIST_SINGLE 0
+#define DIST_BURST  1
+#define DIST_NUM_TYPES 2
+
+struct worker_params {
+   struct rte_distributor *d;
+   struct rte_distributor_burst *db;
+   int dist_type;
+};
+
+struct worker_params worker_params;
+
 /* statics - all zero-initialized by default */
 static volatile int quit;  /**< general quit variable for all threads */
 static volatile int zero_quit; /**< var for when we just want thr0 to quit*/
@@ -81,17 +94,36 @@ static int
 handle_work(void *arg)
 {
struct rte_mbuf *pkt = NULL;
-   struct rte_distributor *d = arg;
-   unsigned count = 0;
-   unsigned id = __sync_fetch_and_add(&worker_idx, 1);
-
-   pkt = rte_distributor_get_pkt(d, id, NULL);
-   while (!quit) {
+   struct rte_mbuf *buf[8] __rte_cache_aligned;
+   struct worker_params *wp = arg;
+   struct rte_distributor *d = wp->d;
+   struct rte_distributor_burst *db = wp->db;
+   unsigned int count = 0, num = 0;
+   unsigned int id = __sync_fetch_and_add(&worker_idx, 1);
+   int i;
+
+   if (wp->dist_type == DIST_SINGLE) {
+   pkt = rte_distributor_get_pkt(d, id, NULL);
+   while (!quit) {
+   worker_stats[id].handled_packets++, count++;
+   pkt = rte_distributor_get_pkt(d, id, pkt);
+   }
worker_stats[id].handled_packets++, count++;
-   pkt = rte_distributor_get_pkt(d, id, pkt);
+   rte_distributor_return_pkt(d, id, pkt);
+   } else {
+   for (i = 0; i < 8; i++)
+   buf[i] = NULL;
+   num = rte_distributor_get_pkt_burst(db, id, buf, buf, num);
+   while (!quit) {
+   worker_stats[id].handled_packets += num;
+   count += num;
+   num = rte_distributor_get_pkt_burst(db, id,
+   buf, buf, num);
+   }
+   worker_stats[id].handled_packets += num;
+   count += num;
+   rte_distributor_return_pkt_burst(db, id, buf, num);
}
-   worker_stats[id].handled_packets++, count++;
-   rte_distributor_return_pkt(d, id, pkt);
return 0;
 }
 
@@ -107,12 +139,21 @@ handle_work(void *arg)
  *   not necessarily in the same order (as different flows).
  */
 static int
-sanity_test(struct rte_distributor *d, struct rte_mempool *p)
+sanity_test(struct worker_params *wp, struct rte_mempool *p)
 {
+   struct rte_distributor *d = wp->d;
+   struct rte_distributor_burst *db = wp->db;
struct rte_mbuf *bufs[BURST];
-   unsigned i;
+   struct rte_mbuf *returns[BURST*2];
+   unsigned int i;
+   unsigned int retries;
+   unsigned int count = 0;
+
+   if (wp->dist_type == DIST_SINGLE)
+   printf("=== Basic distributor sanity tests (single) ===\n");
+   else
+   printf("=== Basic distributor sanity tests (burst) ===\n");
 
-   printf("=== Basic distributor sanity tests ===\n");
clear_packet_count();
if (rte_mempool_get_bulk(p, (void *)bufs, BURST) != 0) {
printf("line %d: Error getting mbufs from pool\n", __LINE__);
@@ -124,8 +165,21 @@ sanity_test(struct rte_distributor *d, struct rte_mempool 
*p)
for (i = 0; i < BURST; i++)
bufs[i]->hash.usr = 0;
 
-   rte_distributor_process(d, bufs, BURST);
-   rte_distributor_flush(d);
+   if (wp->dist_type == DIST_SINGLE) {
+   rte_distributor_process(d, bufs, BURST);
+   rte_distributor_flush(d);
+   } else {
+   rte_distributor_process_burst(db, bufs, BURST);
+   count = 0;
+   do {
+
+   rte_distributor_flush_burst(db);
+   count += rte_distributor_returned_pkts_burst(db,
+   returns, BURST*2);
+   } while (count < BURST);
+   }
+
+
if (total_packet_count() != BURST) {
printf("Line %d: Error, not all packets flushed. "
"Expected %u, got %u\n",
@@ -146,8 +200,18 @@ 

[dpdk-dev] [PATCH v6 5/6] examples/distributor_app: showing burst API

[David Hunt]

Signed-off-by: David Hunt 
---
 examples/distributor/main.c | 511 ++--
 1 file changed, 393 insertions(+), 118 deletions(-)

diff --git a/examples/distributor/main.c b/examples/distributor/main.c
index e7641d2..b0d8b31 100644
--- a/examples/distributor/main.c
+++ b/examples/distributor/main.c
@@ -1,8 +1,7 @@
 /*-
  *   BSD LICENSE
  *
- *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
- *   All rights reserved.
+ *   Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
  *
  *   Redistribution and use in source and binary forms, with or without
  *   modification, are permitted provided that the following conditions
@@ -31,6 +30,8 @@
  *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
+#define BURST_API 1
+
 #include 
 #include 
 #include 
@@ -43,39 +44,87 @@
 #include 
 #include 
 #include 
+#if BURST_API
+#include 
+#else
 #include 
+#endif
 
-#define RX_RING_SIZE 256
-#define TX_RING_SIZE 512
+#define RX_QUEUE_SIZE 512
+#define TX_QUEUE_SIZE 512
 #define NUM_MBUFS ((64*1024)-1)
-#define MBUF_CACHE_SIZE 250
+#define MBUF_CACHE_SIZE 128
+#if BURST_API
+#define BURST_SIZE 64
+#define SCHED_RX_RING_SZ 8192
+#define SCHED_TX_RING_SZ 65536
+#else
 #define BURST_SIZE 32
-#define RTE_RING_SZ 1024
+#define SCHED_RX_RING_SZ 1024
+#define SCHED_TX_RING_SZ 1024
+#endif
+#define BURST_SIZE_TX 32
 
 #define RTE_LOGTYPE_DISTRAPP RTE_LOGTYPE_USER1
 
+#define ANSI_COLOR_RED "\x1b[31m"
+#define ANSI_COLOR_RESET   "\x1b[0m"
+
 /* mask of enabled ports */
 static uint32_t enabled_port_mask;
 volatile uint8_t quit_signal;
 volatile uint8_t quit_signal_rx;
+volatile uint8_t quit_signal_dist;
+volatile uint8_t quit_signal_work;
 
 static volatile struct app_stats {
struct {
uint64_t rx_pkts;
uint64_t returned_pkts;
uint64_t enqueued_pkts;
+   uint64_t enqdrop_pkts;
} rx __rte_cache_aligned;
+   int pad1 __rte_cache_aligned;
+
+   struct {
+   uint64_t in_pkts;
+   uint64_t ret_pkts;
+   uint64_t sent_pkts;
+   uint64_t enqdrop_pkts;
+   } dist __rte_cache_aligned;
+   int pad2 __rte_cache_aligned;
 
struct {
uint64_t dequeue_pkts;
uint64_t tx_pkts;
+   uint64_t enqdrop_pkts;
} tx __rte_cache_aligned;
+   int pad3 __rte_cache_aligned;
+
+   uint64_t worker_pkts[64] __rte_cache_aligned;
+
+   int pad4 __rte_cache_aligned;
+
+   uint64_t worker_bursts[64][8] __rte_cache_aligned;
+
+   int pad5 __rte_cache_aligned;
+
+   uint64_t port_rx_pkts[64] __rte_cache_aligned;
+   uint64_t port_tx_pkts[64] __rte_cache_aligned;
 } app_stats;
 
+struct app_stats prev_app_stats;
+
 static const struct rte_eth_conf port_conf_default = {
.rxmode = {
.mq_mode = ETH_MQ_RX_RSS,
.max_rx_pkt_len = ETHER_MAX_LEN,
+   .split_hdr_size = 0,
+   .header_split   = 0, /**< Header Split disabled */
+   .hw_ip_checksum = 1, /**< IP checksum offload enabled */
+   .hw_vlan_filter = 0, /**< VLAN filtering disabled */
+   .jumbo_frame= 0, /**< Jumbo Frame Support disabled */
+   .hw_strip_crc   = 0, /**< CRC stripped by hardware */
},
.txmode = {
.mq_mode = ETH_MQ_TX_NONE,
@@ -93,6 +142,8 @@ struct output_buffer {
struct rte_mbuf *mbufs[BURST_SIZE];
 };
 
+static void print_stats(void);
+
 /*
  * Initialises a given port using global settings and with the rx buffers
  * coming from the mbuf_pool passed as parameter
@@ -101,9 +152,13 @@ static inline int
 port_init(uint8_t port, struct rte_mempool *mbuf_pool)
 {
struct rte_eth_conf port_conf = port_conf_default;
-   const uint16_t rxRings = 1, txRings = rte_lcore_count() - 1;
-   int retval;
+   const uint16_t rxRings = 1;
+   uint16_t txRings = rte_lcore_count() - 1;
uint16_t q;
+   int retval;
+
+   if (txRings > RTE_MAX_ETHPORTS)
+   txRings = RTE_MAX_ETHPORTS;
 
if (port >= rte_eth_dev_count())
return -1;
@@ -113,7 +168,7 @@ port_init(uint8_t port, struct rte_mempool *mbuf_pool)
return retval;
 
for (q = 0; q < rxRings; q++) {
-   retval = rte_eth_rx_queue_setup(port, q, RX_RING_SIZE,
+   retval = rte_eth_rx_queue_setup(port, q, RX_QUEUE_SIZE,
rte_eth_dev_socket_id(port),
NULL, mbuf_pool);
if (retval < 0)
@@ -121,7 +176,7 @@ port_init(uint8_t port, struct rte_mempool *mbuf_pool)
}
 
for (q = 0; q < txRings; q++) {
-   retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
+   retval = rte_eth_tx_queue_setup(port, q, TX_QUEUE_SIZE,
  

[dpdk-dev] [PATCH v6 6/6] doc: distributor library changes for new burst API

[David Hunt]

Signed-off-by: David Hunt 
---
 doc/guides/prog_guide/packet_distrib_lib.rst | 4 
 1 file changed, 4 insertions(+)

diff --git a/doc/guides/prog_guide/packet_distrib_lib.rst 
b/doc/guides/prog_guide/packet_distrib_lib.rst
index b5bdabb..dffd4ad 100644
--- a/doc/guides/prog_guide/packet_distrib_lib.rst
+++ b/doc/guides/prog_guide/packet_distrib_lib.rst
@@ -42,6 +42,10 @@ The model of operation is shown in the diagram below.
 
Packet Distributor mode of operation
 
+There are two versions of the API in the distributor Library, one which sends 
one packet at a time to workers,
+and another which sends bursts of up to 8 packets at a time to workers. The 
functions names of the second API
+are identified by "_burst", and must not be intermixed with the single packet 
API. The operations described below
+apply to both API's, select which API you wish to use by including the 
relevant header file.
 
 Distributor Core Operation
 --
-- 
2.7.4

[dpdk-dev] [PATCH v6 0/6] distributor library performance enhancements

[David Hunt]

This patch aims to improve the throughput of the distributor library.

It uses a similar handshake mechanism to the previous version of
the library, in that bits are used to indicate when packets are ready
to be sent to a worker and ready to be returned from a worker. One main
difference is that instead of sending one packet in a cache line, it makes
use of the 7 free spaces in the same cache line in order to send up to
8 packets at a time to/from a worker.

The flow matching algorithm has had significant re-work, and now keeps an
array of inflight flows and an array of backlog flows, and matches incoming
flows to the inflight/backlog flows of all workers so that flow pinning to
workers can be maintained.

The Flow Match algorithm has both scalar and a vector versions, and a
function pointer is used to select the post appropriate function at run time,
depending on the presence of the SSE2 cpu flag. On non-x86 platforms, the
the scalar match function is selected, which should still gives a good boost
in performance over the non-burst API.

v2 changes:
  * Created a common distributor_priv.h header file with common
definitions and structures.
  * Added a scalar version so it can be built and used on machines without
sse2 instruction set
  * Added unit autotests
  * Added perf autotest

v3 changes:
  * Addressed mailing list review comments
  * Test code removal
  * Split out SSE match into separate file to facilitate NEON addition
  * Cleaned up conditional compilation flags for SSE2
  * Addressed c99 style compilation errors
  * rebased on latest head (Jan 2 2017, Happy New Year to all)

v4 changes:
   * fixed issue building shared libraries

v5 changes:
   * Removed some un-needed code around retries in worker API calls
   * Cleanup due to review comments on mailing list
   * Cleanup of non-x86 platform compilation, fallback to scalar match

v6 changes:
   * Fixed intermittent segfault where num pkts not divisible
 by BURST_SIZE
   * Cleanup due to review comments on mailing list
   * Renamed _priv.h to _private.h.

Notes:
   Apps must now work in bursts, as up to 8 are given to a worker at a time
   For performance in matching, Flow ID's are 15-bits
   Original API (and code) is kept for backward compatibility

Performance Gains
   2.2GHz Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz
   2 x XL710 40GbE NICS to 2 x 40Gbps traffic generator channels 64b packets
   separate cores for rx, tx, distributor
1 worker  - 4.8x
4 workers - 2.9x
8 workers - 1.8x
   12 workers - 2.1x
   16 workers - 1.8x

[PATCH v6 1/6] lib: distributor performance enhancements
[PATCH v6 2/6] lib: add distributor vector flow matching
[PATCH v6 3/6] test: unit tests for new distributor burst API
[PATCH v6 4/6] test: add distributor perf autotest
[PATCH v6 5/6] examples/distributor_app: showing burst API
[PATCH v6 6/6] doc: distributor library changes for new burst API

Re: [dpdk-dev] [RFC] lib/librte_ether: consistent PMD batching behavior

[Adrien Mazarguil]

On Fri, Jan 20, 2017 at 11:48:22AM +, Bruce Richardson wrote:
> On Fri, Jan 20, 2017 at 11:24:40AM +, Ananyev, Konstantin wrote:
> > > 
> > > From: Andrew Rybchenko [mailto:arybche...@solarflare.com]
> > > Sent: Friday, January 20, 2017 10:26 AM
> > > To: Yang, Zhiyong ; dev@dpdk.org
> > > Cc: thomas.monja...@6wind.com; Richardson, Bruce 
> > > ; Ananyev, Konstantin
> > > 
> > > Subject: Re: [dpdk-dev] [RFC] lib/librte_ether: consistent PMD batching 
> > > behavior
> > > 
> > > On 01/20/2017 12:51 PM, Zhiyong Yang wrote:
> > > The rte_eth_tx_burst() function in the file Rte_ethdev.h is invoked to
> > > transmit output packets on the output queue for DPDK applications as
> > > follows.
> > > 
> > > static inline uint16_t
> > > rte_eth_tx_burst(uint8_t port_id, uint16_t queue_id,
> > >  struct rte_mbuf **tx_pkts, uint16_t nb_pkts);
> > > 
> > > Note: The fourth parameter nb_pkts: The number of packets to transmit.
> > > The rte_eth_tx_burst() function returns the number of packets it actually
> > > sent. The return value equal to *nb_pkts* means that all packets have been
> > > sent, and this is likely to signify that other output packets could be
> > > immediately transmitted again. Applications that implement a "send as many
> > > packets to transmit as possible" policy can check this specific case and
> > > keep invoking the rte_eth_tx_burst() function until a value less than
> > > *nb_pkts* is returned.
> > > 
> > > When you call TX only once in rte_eth_tx_burst, you may get different
> > > behaviors from different PMDs. One problem that every DPDK user has to
> > > face is that they need to take the policy into consideration at the app-
> > > lication level when using any specific PMD to send the packets whether or
> > > not it is necessary, which brings usage complexities and makes DPDK users
> > > easily confused since they have to learn the details on TX function limit
> > > of specific PMDs and have to handle the different return value: the number
> > > of packets transmitted successfully for various PMDs. Some PMDs Tx func-
> > > tions have a limit of sending at most 32 packets for every invoking, some
> > > PMDs have another limit of at most 64 packets once, another ones have imp-
> > > lemented to send as many packets to transmit as possible, etc. This will
> > > easily cause wrong usage for DPDK users.
> > > 
> > > This patch proposes to implement the above policy in DPDK lib in order to
> > > simplify the application implementation and avoid the incorrect invoking
> > > as well. So, DPDK Users don't need to consider the implementation policy
> > > and to write duplicated code at the application level again when sending
> > > packets. In addition to it, the users don't need to know the difference of
> > > specific PMD TX and can transmit the arbitrary number of packets as they
> > > expect when invoking TX API rte_eth_tx_burst, then check the return value
> > > to get the number of packets actually sent.
> > > 
> > > How to implement the policy in DPDK lib? Two solutions are proposed below.
> > > 
> > > Solution 1:
> > > Implement the wrapper functions to remove some limits for each specific
> > > PMDs as i40e_xmit_pkts_simple and ixgbe_xmit_pkts_simple do like that.
> > > 
> > > > IMHO, the solution is a bit better since it:
> > > > 1. Does not affect other PMDs at all
> > > > 2. Could be a bit faster for the PMDs which require it since has no 
> > > >indirect
> > > >    function call on each iteration
> > > > 3. No ABI change
> > 
> > I also would prefer solution number 1 for the reasons outlined by Andrew 
> > above.
> > Also, IMO current limitation for number of packets to TX in some Intel PMD 
> > TX routines
> > are sort of artificial:
> > - they are not caused by any real HW limitations
> > - avoiding them at PMD level shouldn't cause any performance or functional 
> > degradation.
> > So I don't see any good reason why instead of fixing these limitations in
> > our own PMDs we are trying to push them to the upper (rte_ethdev) layer.

For what it's worth, I agree with Konstantin. Wrappers should be as thin as
possible on top of PMD functions, they are not helpers. We could define a
set of higher level functions for this purpose though.

In the meantime, exposing and documenting PMD limitations seems safe enough.

We could assert that RX/TX burst requests larger than the size of the target
queue are unlikely to be fully met (i.e. PMDs usually do not check for
completions in the middle of a TX burst).

> > Konstantin
> > 
> The main advantage I see is that it should make it a bit easier for
> driver writers, since they have a tighter set of constraints to work
> with for packet RX and Tx. The routines only have to handle requests for
> packets in the range 0-N, rather than not having an upper bound on the
> request. It also then saves code duplicating with having multiple
> drivers having the same outer-loop code for handling arbitrarily large
> requests.
> 
> No big

Re: [dpdk-dev] [PATCH v6 0/6] distributor library performance enhancements

[Bruce Richardson]

On Mon, Jan 23, 2017 at 09:24:34AM +, David Hunt wrote:
> This patch aims to improve the throughput of the distributor library.
> 
> It uses a similar handshake mechanism to the previous version of
> the library, in that bits are used to indicate when packets are ready
> to be sent to a worker and ready to be returned from a worker. One main
> difference is that instead of sending one packet in a cache line, it makes
> use of the 7 free spaces in the same cache line in order to send up to
> 8 packets at a time to/from a worker.
> 
> The flow matching algorithm has had significant re-work, and now keeps an
> array of inflight flows and an array of backlog flows, and matches incoming
> flows to the inflight/backlog flows of all workers so that flow pinning to
> workers can be maintained.
> 
> The Flow Match algorithm has both scalar and a vector versions, and a
> function pointer is used to select the post appropriate function at run time,
> depending on the presence of the SSE2 cpu flag. On non-x86 platforms, the
> the scalar match function is selected, which should still gives a good boost
> in performance over the non-burst API.
> 
> v2 changes:
>   * Created a common distributor_priv.h header file with common
> definitions and structures.
>   * Added a scalar version so it can be built and used on machines without
> sse2 instruction set
>   * Added unit autotests
>   * Added perf autotest
> 
> v3 changes:
>   * Addressed mailing list review comments
>   * Test code removal
>   * Split out SSE match into separate file to facilitate NEON addition
>   * Cleaned up conditional compilation flags for SSE2
>   * Addressed c99 style compilation errors
>   * rebased on latest head (Jan 2 2017, Happy New Year to all)
> 
> v4 changes:
>* fixed issue building shared libraries
> 
> v5 changes:
>* Removed some un-needed code around retries in worker API calls
>* Cleanup due to review comments on mailing list
>* Cleanup of non-x86 platform compilation, fallback to scalar match
> 
> v6 changes:
>* Fixed intermittent segfault where num pkts not divisible
>  by BURST_SIZE
>* Cleanup due to review comments on mailing list
>* Renamed _priv.h to _private.h.
> 
> Notes:
>Apps must now work in bursts, as up to 8 are given to a worker at a time
>For performance in matching, Flow ID's are 15-bits
>Original API (and code) is kept for backward compatibility
> 
> Performance Gains
>2.2GHz Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz
>2 x XL710 40GbE NICS to 2 x 40Gbps traffic generator channels 64b packets
>separate cores for rx, tx, distributor
> 1 worker  - 4.8x
> 4 workers - 2.9x
> 8 workers - 1.8x
>12 workers - 2.1x
>16 workers - 1.8x
> 

Series Acked-by: Bruce Richardson 

[dpdk-dev] [PATCH] mempool: fix stack handler dequeue

[Olivier Matz]

The return value of the stack handler is wrong: it should be 0 on
success, not the number of objects dequeued.

This could lead to memory leaks depending on how the caller checks the
return value (ret < 0 or ret != 0). This was also breaking autotests
with debug enabled, because the debug cookies are only updated when the
function returns 0, so the cookies were not updated, leading to
an abort().

Fixes: 295a530b0844 ("mempool: add stack mempool handler")

CC: sta...@dpdk.org
Signed-off-by: Olivier Matz 
---
 lib/librte_mempool/rte_mempool_stack.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/librte_mempool/rte_mempool_stack.c 
b/lib/librte_mempool/rte_mempool_stack.c
index 5fd8af2..817f77e 100644
--- a/lib/librte_mempool/rte_mempool_stack.c
+++ b/lib/librte_mempool/rte_mempool_stack.c
@@ -118,7 +118,7 @@ stack_dequeue(struct rte_mempool *mp, void **obj_table,
 
s->len -= n;
rte_spinlock_unlock(&s->sl);
-   return n;
+   return 0;
 }
 
 static unsigned
-- 
2.8.1

Re: [dpdk-dev] [RFC] Add GRO support in DPDK

[Stephen Hemminger]

On Mon, 23 Jan 2017 21:03:12 +0800
Jiayu Hu  wrote:

> With the support of hardware segmentation techniques in DPDK, the
> networking stack overheads of send-side of applications, which directly
> leverage DPDK, have been greatly reduced. But for receive-side, numbers of
> segmented packets seriously burden the networking stack of applications.
> Generic Receive Offload (GRO) is a widely used method to solve the
> receive-side issue, which gains performance by reducing the amount of
> packets processed by the networking stack. But currently, DPDK doesn't
> support GRO. Therefore, we propose to add GRO support in DPDK, and this
> RFC is used to explain the basic DPDK GRO design.
> 
> DPDK GRO is a SW-based packets assembly library, which provides GRO
> abilities for numbers of protocols. In DPDK GRO, packets are merged
> before returning to applications and after receiving from drivers.
> 
> In DPDK, GRO is a capability of NIC drivers. That support GRO or not and
> what GRO types are supported are up to NIC drivers. Different drivers may
> support different GRO types. By default, drivers enable all supported GRO
> types. For applications, they can inquire the supported GRO types by
> each driver, and can control what GRO types are applied. For example,
> ixgbe supports TCP and UDP GRO, but the application just needs TCP GRO.
> The application can disable ixgbe UDP GRO.
> 
> To support GRO, a driver should provide a way to tell applications what
> GRO types are supported, and provides a GRO function, which is in charge
> of assembling packets. Since different drivers may support different GRO
> types, their GRO functions may be different. For applications, they don't
> need extra operations to enable GRO. But if there are some GRO types that
> are not needed, applications can use an API, like
> rte_eth_gro_disable_protocols, to disable them. Besides, they can
> re-enable the disabled ones.
> 
> The GRO function processes numbers of packets at a time. In each
> invocation, what GRO types are applied depends on applications, and the
> amount of packets to merge depends on the networking status and
> applications. Specifically, applications determine the maximum number of
> packets to be processed by the GRO function, but how many packets are
> actually processed depends on if there are available packets to receive.
> For example, the receive-side application asks the GRO function to
> process 64 packets, but the sender only sends 40 packets. At this time,
> the GRO function returns after processing 40 packets. To reassemble the
> given packets, the GRO function performs an "assembly procedure" on each
> packet. We use an example to demonstrate this procedure. Supposing the
> GRO function is going to process packetX, it will do the following two
> things:
>   a. Find a L4 assembly function according to the packet type of
>   packetX. A L4 assembly function is in charge of merging packets of a
>   specific type. For example, TCPv4 assembly function merges packets
>   whose L3 IPv4 and L4 is TCP. Each L4 assembly function has a packet
>   array, which keeps the packets that are unable to assemble.
>   Initially, the packet array is empty;
>   b. The L4 assembly function traverses own packet array to find a
>   mergeable packet (comparing Ethernet, IP and L4 header fields). If
>   finds, merges it and packetX via chaining them together; if doesn't,
>   allocates a new array element to store packetX and updates element
>   number of the array.
> After performing the assembly procedure to all packets, the GRO function
> combines the results of all packet arrays, and returns these packets to
> applications.
> 
> There are lots of ways to implement the above design in DPDK. One of the
> ways is:
>   a. Drivers tell applications what GRO types are supported via
>   dev->dev_ops->dev_infos_get;
>   b. When initialize, drivers register own GRO function as a RX
>   callback, which is invoked inside rte_eth_rx_burst. The name of the
>   GRO function should be like xxx_gro_receive (e.g. ixgbe_gro_receive).
>   Currently, the RX callback can only process the packets returned by
>   dev->rx_pkt_burst each time, and the maximum packet number
>   dev->rx_pkt_burst returns is determined by each driver, which can't
>   be interfered by applications. Therefore, to implement the above GRO
>   design, we have to modify current RX implementation to make driver
>   return packets as many as possible until the packet number meets the
>   demand of applications or there are not available packets to receive.
>   This modification is also proposed in patch:
>   http://dpdk.org/ml/archives/dev/2017-January/055887.html;
>   c. The GRO types to apply and the maximum number of packets to merge
>   are passed by resetting RX callback parameters. It can be achieved by
>   invoking rte_eth_rx_callback;
>   d. Simply, we can just st

Re: [dpdk-dev] [PATCH] mk: parallelize make config

[Olivier Matz]

Hi Ferruh,

On Sun, 22 Jan 2017 01:50:34 +, Ferruh Yigit
 wrote:
> make config dependency resolving was always running serial,
> parallelize it for better performance.
> 
> $ time make T=x86_64-native-linuxapp-gcc config
> real0m12.633s
> 
> $ time make -j8 T=x86_64-native-linuxapp-gcc config
> real0m1.826s
> 
> Signed-off-by: Ferruh Yigit 

I have a patch that fix the same issue (configuration takes to long),
but done differently. It is more intrusive, since it rework the way
DEPDIRS are used, but it does not require to use -j.

I'm sending it as a reply to this thread.

Regards,
Olivier

[dpdk-dev] [PATCH] mk: optimize directory dependencies

[Olivier Matz]

Before this patch, the management of dependencies between directories
had several issues:

- the generation of .depdirs, done at configuration is slow: it can take
  more than one minute on some slow targets (usually ~10s on a standard
  PC).

- for instance, it is possible to expressed a dependency like:
  - app/foo depends on lib/librte_foo
  - and lib/librte_foo depends on app/bar
  But this won't work because the directories are traversed with a
  depth-first algorithm, so we have to choose between doing 'app' before
  or after 'lib'.

- the script depdirs-rule.sh is too complex.

- we cannot use "make -d" for debug, because the output of make is used for
  the generation of .depdirs.

This patch moves the DEPDIRS-* variables in the upper Makefile, making
the dependencies much easier to calculate. A DEPDIRS variable is still
used to process library dependencies in LDLIBS.

After this commit, "make config" is almost immediate.

Signed-off-by: Olivier Matz 
---
 MAINTAINERS  |  1 -
 app/cmdline_test/Makefile|  3 -
 app/pdump/Makefile   |  1 -
 app/proc_info/Makefile   |  3 -
 app/test-acl/Makefile|  1 -
 app/test-pipeline/Makefile   |  3 -
 app/test-pmd/Makefile|  1 -
 app/test/Makefile|  3 -
 buildtools/depdirs-rule.sh   | 95 
 buildtools/pmdinfogen/Makefile   |  2 -
 doc/guides/prog_guide/dev_kit_build_system.rst   |  2 +-
 doc/guides/prog_guide/dev_kit_root_make_help.rst | 28 ---
 drivers/crypto/Makefile  | 12 +++
 drivers/crypto/aesni_gcm/Makefile|  7 --
 drivers/crypto/aesni_mb/Makefile |  7 --
 drivers/crypto/armv8/Makefile|  7 --
 drivers/crypto/kasumi/Makefile   |  7 --
 drivers/crypto/null/Makefile |  7 --
 drivers/crypto/openssl/Makefile  |  7 --
 drivers/crypto/qat/Makefile  |  7 --
 drivers/crypto/snow3g/Makefile   |  7 --
 drivers/crypto/zuc/Makefile  |  7 --
 drivers/net/Makefile | 31 
 drivers/net/af_packet/Makefile   |  7 --
 drivers/net/bnx2x/Makefile   |  4 -
 drivers/net/bnxt/Makefile|  6 --
 drivers/net/bonding/Makefile |  9 ---
 drivers/net/cxgbe/Makefile   |  5 --
 drivers/net/e1000/Makefile   |  5 --
 drivers/net/ena/Makefile |  5 --
 drivers/net/enic/Makefile|  6 --
 drivers/net/fm10k/Makefile   |  6 --
 drivers/net/i40e/Makefile|  7 --
 drivers/net/ixgbe/Makefile   |  6 --
 drivers/net/mlx4/Makefile|  6 --
 drivers/net/mlx5/Makefile|  7 --
 drivers/net/mpipe/Makefile   |  3 -
 drivers/net/nfp/Makefile |  5 --
 drivers/net/null/Makefile|  7 --
 drivers/net/pcap/Makefile|  7 --
 drivers/net/qede/Makefile|  5 --
 drivers/net/ring/Makefile|  5 --
 drivers/net/sfc/Makefile |  9 ---
 drivers/net/szedata2/Makefile|  7 --
 drivers/net/tap/Makefile |  7 --
 drivers/net/thunderx/Makefile|  4 -
 drivers/net/vhost/Makefile   |  8 --
 drivers/net/virtio/Makefile  |  6 --
 drivers/net/vmxnet3/Makefile |  5 --
 drivers/net/xenvirt/Makefile |  6 --
 examples/ethtool/Makefile|  3 +
 examples/ethtool/lib/Makefile|  4 -
 lib/Makefile | 35 +
 lib/librte_acl/Makefile  |  3 -
 lib/librte_cfgfile/Makefile  |  3 -
 lib/librte_cmdline/Makefile  |  3 -
 lib/librte_cryptodev/Makefile|  7 --
 lib/librte_distributor/Makefile  |  4 -
 lib/librte_eal/Makefile  |  3 +
 lib/librte_eal/bsdapp/eal/Makefile   |  3 -
 lib/librte_eal/linuxapp/Makefile |  3 +
 lib/librte_eal/linuxapp/eal/Makefile |  3 -
 lib/librte_eal/linuxapp/kni/Makefile |  3 -
 lib/librte_eal/linuxapp/xen_dom0/Makefile|  3 -
 lib/librte_efd/Makefile  |  5 --
 lib/librte_ether/Makefile|  3 -
 lib/librte_hash/Makefile |  3 -
 lib/librte_ip_frag/Makefile  |  6 --
 lib/librte_jobstats/Makefile |  3 -
 li

Re: [dpdk-dev] [PATCHv6 02/33] drivers/common/dpaa2: adding qbman driver

[Ferruh Yigit]

On 1/23/2017 11:59 AM, Hemant Agrawal wrote:
> QBMAN, is a hardware block which interfaces with the other
> accelerating hardware blocks (For e.g., WRIOP) on NXP's DPAA2
> SoC for queue, buffer and packet scheduling.
> 
> This patch introduces a userspace driver for interfacing with
> the QBMAN hw block.
> 
> The qbman-portal component provides APIs to do the low level
> hardware bit twiddling for operations such as:
>   -initializing Qman software portals
>   -building and sending portal commands
>   -portal interrupt configuration and processing
> 
> This same/similar code is used in kernel and compat file is used
> to make it working in user space.
> 
> Signed-off-by: Geoff Thorpe 
> Signed-off-by: Roy Pledge 
> Signed-off-by: Hemant Agrawal 
> ---
<...>

> --- a/config/common_base
> +++ b/config/common_base
> @@ -287,7 +287,6 @@ CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_TX=n
>  CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_DRIVER=n
>  CONFIG_RTE_LIBRTE_THUNDERX_NICVF_DEBUG_MBOX=n
>  
> -#

Minor typo ..

>  # Compile burst-oriented VIRTIO PMD driver
>  #
>  CONFIG_RTE_LIBRTE_VIRTIO_PMD=y

<...>

> --- /dev/null
> +++ b/drivers/common/dpaa2/qbman/rte_common_dpaa2_qbman_version.map
> @@ -0,0 +1,27 @@
> +DPDK_17.02 {
> + global:
> +
> + qbman_check_command_complete;
> + qbman_eq_desc_clear;
> + qbman_eq_desc_set_fq;
> + qbman_eq_desc_set_no_orp;
> + qbman_eq_desc_set_qd;
> + qbman_eq_desc_set_response;
> + qbman_get_version;
> + qbman_pull_desc_clear;
> + qbman_pull_desc_set_fq;
> + qbman_pull_desc_set_numframes;
> + qbman_pull_desc_set_storage;
> + qbman_release_desc_clear;
> + qbman_release_desc_set_bpid;
> + qbman_result_DQ_fd;
> + qbman_result_DQ_flags;
> + qbman_result_has_new_result;
> + qbman_swp_acquire;
> + qbman_swp_init;
> + qbman_swp_pull;
> + qbman_swp_release;
> + qbman_swp_send_multiple;

Overall, dpdk library exported APIs not having DPDK prefix (rte_) is a
concern, which already pointed by Thomas.

I guess only user of this library will be other dpaa2 code, so these are
not really APIs. Not sure how to proceed.
I think I have seen "_rte" prefix used in some APIs to say that is
internal API, does it make sense to use that API here?

> +
> + local: *;
> +};
> 

Re: [dpdk-dev] [PATCHv6 12/33] net/dpaa2: introducing NXP dpaa2 pmd driver

[Ferruh Yigit]

On 1/23/2017 11:59 AM, Hemant Agrawal wrote:
> add support for fsl-mc bus based dpaa2 pmd driver.
> 
> Signed-off-by: Hemant Agrawal 
> ---
<...>

> diff --git a/drivers/net/dpaa2/Makefile b/drivers/net/dpaa2/Makefile
> new file mode 100644
> index 000..f85aa9f
> --- /dev/null
> +++ b/drivers/net/dpaa2/Makefile
<...>
> +
> +# library dependencies
> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_eal lib/librte_ether
> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_bus_fslmc

Is this dependency correct?

> +
> +include $(RTE_SDK)/mk/rte.lib.mk

<...>

Re: [dpdk-dev] [PATCHv6 16/33] drivers/pool/dpaa2: adding hw offloaded mempool

[Ferruh Yigit]

On 1/23/2017 11:59 AM, Hemant Agrawal wrote:
> Adding NXP DPAA2 architecture specific mempool support
> Each mempool instance is represented by a DPBP object
> from the FSL-MC bus.
> 
> This patch also registers a dpaa2 type MEMPOOL OPS
> 
> Signed-off-by: Hemant Agrawal 
> ---
<...>

> diff --git a/drivers/common/Makefile b/drivers/common/Makefile
> index b52931c..0bb75b5 100644
> --- a/drivers/common/Makefile
> +++ b/drivers/common/Makefile
> @@ -35,7 +35,11 @@ ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_PMD),y)
>  CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_DPAA2_PMD)
>  endif
>  
> -ifeq ($(CONFIG_RTE_LIBRTE_FSLMC_BUS),y)
> +ifeq ($(CONFIG_RTE_LIBRTE_DPAA2_POOL),y)
> +CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_DPAA2_POOL)
> +endif
> +
> +ifneq ($(CONFIG_RTE_LIBRTE_FSLMC_BUS),y)

I guess this is a typo, but this prevents DPAA2_COMMON to be compiled !!

>  CONFIG_RTE_LIBRTE_DPAA2_COMMON = $(CONFIG_RTE_LIBRTE_FSLMC_BUS)
>  endif
>  

<...>
> +# library dependencies
> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_POOL) += lib/librte_eal
> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_POOL) += lib/librte_mempool
> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_POOL) += lib/librte_common_dpaa2_qbman

This dependeny doesn not looks correct, there is no folder like that.

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_POOL) += lib/librte_bus_fslmc
> +
> +include $(RTE_SDK)/mk/rte.lib.mk

<...>

Re: [dpdk-dev] [PATCHv6 18/33] net/dpaa2: adding eth ops to dpaa2

[Ferruh Yigit]

On 1/23/2017 11:59 AM, Hemant Agrawal wrote:
> Signed-off-by: Hemant Agrawal 
> ---
<...>

>  static int
>  dpaa2_dev_init(struct rte_eth_dev *eth_dev)
>  {
<...>
> + eth_dev->data->nb_rx_queues = priv->nb_rx_queues;
> + eth_dev->data->nb_tx_queues = priv->nb_tx_queues;

No need to assign these during init(), app should set these via
rte_eth_dev_configure() API.

> +
> + priv->hw = dpni_dev;
> + priv->hw_id = hw_id;
> + priv->flags = 0;

<...>

Re: [dpdk-dev] [PATCHv6 22/33] net/dpaa2: add support for l3 and l4 checksum offload

[Ferruh Yigit]

On 1/23/2017 11:59 AM, Hemant Agrawal wrote:
> Signed-off-by: Hemant Agrawal 
> ---
<...>
> --- a/drivers/net/dpaa2/Makefile
> +++ b/drivers/net/dpaa2/Makefile
> @@ -66,6 +66,6 @@ DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_eal 
> lib/librte_ether
>  DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_mempool lib/librte_mbuf
>  DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_common_dpaa2_qbman
>  DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_bus_fslmc
> -DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_pmd_dpaa2_pool
> +DEPDIRS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += lib/librte_pool_dpaa2

I understand these are added since DEPDIRS converted to LDLIBS, but this
does not looks right, since these variables mainly dependency solving
and the value provided is not correct for DEPDIRS.

Did you tried adding as "LDLIBS += xx", not tested, but may work.

>  
>  include $(RTE_SDK)/mk/rte.lib.mk
> diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c 
> b/drivers/net/dpaa2/dpaa2_ethdev.c
> index 6de571a..2298246 100644
> --- a/drivers/net/dpaa2/dpaa2_ethdev.c
> +++ b/drivers/net/dpaa2/dpaa2_ethdev.c
> @@ -68,7 +68,17 @@
>   dev_info->min_rx_bufsize = DPAA2_MIN_RX_BUF_SIZE;
>   dev_info->max_rx_queues = (uint16_t)priv->nb_rx_queues;
>   dev_info->max_tx_queues = (uint16_t)priv->nb_tx_queues;
> -
> + dev_info->rx_offload_capa =
> + DEV_RX_OFFLOAD_IPV4_CKSUM |
> + DEV_RX_OFFLOAD_UDP_CKSUM |
> + DEV_RX_OFFLOAD_TCP_CKSUM |
> + DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM;

Is there capabilities always enabled for all devices this driver
support? Or should these be read from some device registers?

> + dev_info->tx_offload_capa =
> + DEV_TX_OFFLOAD_IPV4_CKSUM |
> + DEV_TX_OFFLOAD_UDP_CKSUM |
> + DEV_TX_OFFLOAD_TCP_CKSUM |
> + DEV_TX_OFFLOAD_SCTP_CKSUM |
> + DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM;
>   dev_info->speed_capa = ETH_LINK_SPEED_1G |
>   ETH_LINK_SPEED_2_5G |
>   ETH_LINK_SPEED_10G;
<...>