[dpdk-dev] [PATCH 2/2] lib/librte_eal: Remove unnecessary hugepage zero-filling

[Thomas Monjalon]

2015-11-24 14:44, Stephen Hemminger:
> On Tue, 24 Nov 2015 22:13:28 +0100
> Thomas Monjalon  wrote:
> 
> > 2015-11-22 18:28, Stephen Hemminger:
> > > On Sun, 22 Nov 2015 14:13:35 -0500
> > > Zhihong Wang  wrote:
> > > 
> > > > The kernel fills new allocated (huge) pages with zeros.
> > > > DPDK just has to populate page tables to trigger the allocation.
> > > > 
> > > > Signed-off-by: Zhihong Wang 
> > > 
> > > Nice, especially on slow machines or with large memory.
> > > 
> > > Acked-by: Stephen Hemminger 
> > 
> > Yes very nice.
> > I think it's too late to integrate this change which can have some
> > unpredictable side effects.
> > Do you agree to wait for 2.3?
> 
> What side effects? Either it is zero or it is not.
> Only some broken architecture would have an issue.

I mean it changes the memory allocator behaviour. It's not something we
want to discover a new bug just before the release.
This kind of important change must be integrated at the beginning of the
release cycle.
I'm asking for opinions because it would be really nice to have.

[dpdk-dev] [PATCH v2 00/10] rte_sched: enhancements and cleanups

[Thomas Monjalon]

2015-11-13 09:58, Stephen Hemminger:
> Hierarchal scheduler changes (resend)
> 
> Most of these are cleanups for existing code to make it more compatiable
> with coding style, and eliminate #ifdefs.
> 
> The only substantive change is to allow more subports per scheduler
> instance. This was submitted for DPDK 2.1 but since it required some
> ABI finesse to change; the real part of this was held off for 2.2.
> 
> Stephen Hemminger (10):
>   qos: drop deprecated port hierarchy structure
>   qos: cleanup comments
>   qos: make debugging configurable
>   qos: drop debug #ifdef's for credit check
>   qos: remove debug conditional code around ENQUEUE
>   qos: drop RTE_SCHED_WRR #define
>   qos: cleanup defined constants
>   qos: allow enabling SSE optimizations in config
>   sched: fix coding style
>   sched: allow more subports

Applied with deprecation notice removed, thanks.

What about this change announced as deprecation notice?
+* The scheduler statistics structure will change to allow keeping track of
+  RED actions.

[dpdk-dev] [PATCH v7 05/10] cryptodev: Initial DPDK Crypto APIs and device framework release

[Thomas Monjalon]

2015-11-13 18:58, Declan Doherty:
>  lib/librte_cryptodev/rte_crypto.h  |  613 +
>  lib/librte_cryptodev/rte_cryptodev.c   | 1092 
> 
>  lib/librte_cryptodev/rte_cryptodev.h   |  649 ++

Doxygen reports some errors:

lib/librte_cryptodev/rte_crypto.h:565: warning: unable to resolve reference to 
`rte_crypto_hash_setup_data' for \ref command

lib/librte_cryptodev/rte_crypto.h:585: warning: argument 'm' of command @param 
is not found in the argument list of __rte_crypto_op_reset(struct rte_crypto_op 
*op)

lib/librte_cryptodev/rte_crypto.h:589: warning: unable to resolve reference to 
`rte_crypto_hash_params' for \ref command   
   

lib/librte_cryptodev/rte_crypto.h:593: warning: The following parameters of 
__rte_crypto_op_reset(struct rte_crypto_op *op) are not documented:

lib/librte_cryptodev/rte_cryptodev.h:241: warning: argument 'nb_qp_queue' of 
command @param is not found in the argument list of 
rte_cryptodev_configure(uint8_t dev_id, struct rte_cryptodev_config *config)

lib/librte_cryptodev/rte_cryptodev.h:256: warning: The following parameters of 
rte_cryptodev_configure(uint8_t dev_id, struct rte_cryptodev_config *config) 
are not documented:

lib/librte_cryptodev/rte_cryptodev.h:563: warning: argument 'queue_id' of 
command @param is not found in the argument list of 
rte_cryptodev_enqueue_burst(uint8_t dev_id, uint16_t qp_id, struct rte_mbuf 
**pkts, uint16_t nb_pkts)

lib/librte_cryptodev/rte_cryptodev.h:563: warning: argument 'tx_pkts' of 
command @param is not found in the argument list of 
rte_cryptodev_enqueue_burst(uint8_t dev_id, uint16_t qp_id, struct rte_mbuf 
**pkts, uint16_t nb_pkts)

lib/librte_cryptodev/rte_cryptodev.h:593: warning: The following parameters of 
rte_cryptodev_enqueue_burst(uint8_t dev_id, uint16_t qp_id, struct rte_mbuf 
**pkts, uint16_t nb_pkts) are not documented:

[dpdk-dev] [PATCH] sched: remove unused debug functions

[Thomas Monjalon]

clang warns about unused functions when debug is enabled.

Signed-off-by: Thomas Monjalon 
---
 lib/librte_sched/rte_sched.c | 31 ---
 1 file changed, 31 deletions(-)

diff --git a/lib/librte_sched/rte_sched.c b/lib/librte_sched/rte_sched.c
index 9f0b458..d47cfc2 100644
--- a/lib/librte_sched/rte_sched.c
+++ b/lib/librte_sched/rte_sched.c
@@ -1055,16 +1055,6 @@ rte_sched_port_queue_is_empty(struct rte_sched_port 
*port, uint32_t qindex)
return (queue->qr == queue->qw);
 }

-static inline int
-rte_sched_port_queue_is_full(struct rte_sched_port *port, uint32_t qindex)
-{
-   struct rte_sched_queue *queue = port->queue + qindex;
-   uint16_t qsize = rte_sched_port_qsize(port, qindex);
-   uint16_t qlen = queue->qw - queue->qr;
-
-   return (qlen >= qsize);
-}
-
 #endif /* RTE_SCHED_DEBUG */

 #ifdef RTE_SCHED_COLLECT_STATS
@@ -1156,27 +1146,6 @@ rte_sched_port_set_queue_empty_timestamp(struct 
rte_sched_port *port, uint32_t q

 #ifdef RTE_SCHED_DEBUG

-static inline int
-debug_pipe_is_empty(struct rte_sched_port *port, uint32_t pindex)
-{
-   uint32_t qindex, i;
-
-   qindex = pindex << 4;
-
-   for (i = 0; i < 16; i++) {
-   uint32_t queue_empty = rte_sched_port_queue_is_empty(port, 
qindex + i);
-   uint32_t bmp_bit_clear = (rte_bitmap_get(port->bmp, qindex + i) 
== 0);
-
-   if (queue_empty != bmp_bit_clear)
-   rte_panic("Queue status mismatch for queue %u of pipe 
%u\n", i, pindex);
-
-   if (!queue_empty)
-   return 0;
-   }
-
-   return 1;
-}
-
 static inline void
 debug_check_queue_slab(struct rte_sched_port *port, uint32_t bmp_pos,
   uint64_t bmp_slab)
-- 
2.5.2

[dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Thomas Monjalon]

2015-11-13 18:58, Declan Doherty:
> +Crypto Device Drivers
> +

It is a quite long underlining :)

It raise a question to me:
John, have you reviewed the crypto docs?

[dpdk-dev] [PATCH v7 10/10] l2fwd-crypto: crypto

[Thomas Monjalon]

2015-11-13 18:58, Declan Doherty:
> +   printf("\nStatistics for cryptodev %lu 
> -"
> +  "\nPackets enqueued: %28"PRIu64
> +  "\nPackets dequeued: %28"PRIu64
> +  "\nPackets errors: %30"PRIu64,
> +  cdevid,
> +  crypto_statistics[cdevid].enqueued,
> +  crypto_statistics[cdevid].dequeued,
> +  crypto_statistics[cdevid].errors);

There is a compilation error on 32-bit:

examples/l2fwd-crypto/main.c:252:10:
error: format ?%lu? expects argument of type ?long unsigned int?,
but argument 2 has type ?uint64_t {aka long long unsigned int}?

[dpdk-dev] [PATCH] virtio: fix rx ring descriptor starvation

[Yuanhan Liu]

On Tue, Nov 24, 2015 at 10:20:22PM +0100, Thomas Monjalon wrote:
> Any review, please?

Huawei, would you review it? Sorry that I've not read too much
code about virtio PMD driver yet.

--yliu


> 2015-11-13 09:30, Tom Kiely:
> > If all rx descriptors are processed while transient
> > mbuf exhaustion is present, the rx ring ends up with
> > no available descriptors. Thus no packets are received
> > on that ring. Since descriptor refill is performed post
> > rx descriptor processing, in this case no refill is
> > ever subsequently performed resulting in permanent rx
> > traffic drop.
> > 
> > Signed-off-by: Tom Kiely 

[dpdk-dev] [PATCH 2/2] lib/librte_eal: Remove unnecessary hugepage zero-filling

[Yuanhan Liu]

On Wed, Nov 25, 2015 at 12:04:16AM +0100, Thomas Monjalon wrote:
> 2015-11-24 14:44, Stephen Hemminger:
> > On Tue, 24 Nov 2015 22:13:28 +0100
> > Thomas Monjalon  wrote:
> > 
> > > 2015-11-22 18:28, Stephen Hemminger:
> > > > On Sun, 22 Nov 2015 14:13:35 -0500
> > > > Zhihong Wang  wrote:
> > > > 
> > > > > The kernel fills new allocated (huge) pages with zeros.
> > > > > DPDK just has to populate page tables to trigger the allocation.
> > > > > 
> > > > > Signed-off-by: Zhihong Wang 
> > > > 
> > > > Nice, especially on slow machines or with large memory.
> > > > 
> > > > Acked-by: Stephen Hemminger 
> > > 
> > > Yes very nice.
> > > I think it's too late to integrate this change which can have some
> > > unpredictable side effects.
> > > Do you agree to wait for 2.3?
> > 
> > What side effects? Either it is zero or it is not.
> > Only some broken architecture would have an issue.
> 
> I mean it changes the memory allocator behaviour. It's not something we
> want to discover a new bug just before the release.
> This kind of important change must be integrated at the beginning of the
> release cycle.

+ 1

And it could be a new feature (or highlight) of 2.3: reduced dpdk
startup time by ... :)

--yliu

[dpdk-dev] [PATCH 2/2] lib/librte_eal: Remove unnecessary hugepage zero-filling

[Wang, Zhihong]

> -Original Message-
> From: Thomas Monjalon [mailto:thomas.monjalon at 6wind.com]
> Sent: Wednesday, November 25, 2015 7:04 AM
> To: Stephen Hemminger 
> Cc: Wang, Zhihong ; dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH 2/2] lib/librte_eal: Remove unnecessary
> hugepage zero-filling
> 
> 2015-11-24 14:44, Stephen Hemminger:
> > On Tue, 24 Nov 2015 22:13:28 +0100
> > Thomas Monjalon  wrote:
> >
> > > 2015-11-22 18:28, Stephen Hemminger:
> > > > On Sun, 22 Nov 2015 14:13:35 -0500 Zhihong Wang
> > > >  wrote:
> > > >
> > > > > The kernel fills new allocated (huge) pages with zeros.
> > > > > DPDK just has to populate page tables to trigger the allocation.
> > > > >
> > > > > Signed-off-by: Zhihong Wang 
> > > >
> > > > Nice, especially on slow machines or with large memory.
> > > >
> > > > Acked-by: Stephen Hemminger 
> > >
> > > Yes very nice.
> > > I think it's too late to integrate this change which can have some
> > > unpredictable side effects.
> > > Do you agree to wait for 2.3?
> >
> > What side effects? Either it is zero or it is not.
> > Only some broken architecture would have an issue.
> 
> I mean it changes the memory allocator behaviour. It's not something we want 
> to
> discover a new bug just before the release.
> This kind of important change must be integrated at the beginning of the 
> release
> cycle.
> I'm asking for opinions because it would be really nice to have.

Literally this patch doesn't change anything, it just keeps DPDK from 
zero-filling pages again which have just been zero-filled.
It would be nice to have this patch in DPDK 2.2 since it can reduce the startup 
time nearly by half for hugepage cases.
But I understand longer merge/test window make it safer for a release.
It makes sense either way.

[dpdk-dev] [PATCH] virtio: fix rx ring descriptor starvation

[Xie, Huawei]

On 11/25/2015 9:47 AM, Yuanhan Liu wrote:
> On Tue, Nov 24, 2015 at 10:20:22PM +0100, Thomas Monjalon wrote:
>> Any review, please?
> Huawei, would you review it? Sorry that I've not read too much
> code about virtio PMD driver yet.
Np. will do it by end of this week.
>
>   --yliu
>
>
>> 2015-11-13 09:30, Tom Kiely:
>>> If all rx descriptors are processed while transient
>>> mbuf exhaustion is present, the rx ring ends up with
>>> no available descriptors. Thus no packets are received
>>> on that ring. Since descriptor refill is performed post
>>> rx descriptor processing, in this case no refill is
>>> ever subsequently performed resulting in permanent rx
>>> traffic drop.
>>>
>>> Signed-off-by: Tom Kiely 

[dpdk-dev] no hugepage with UIO poll-mode driver

[Younghwan Go]

Hello,

I'm trying to create rte mempool with my own memory scheme (shared virtual 
memory between CPU & GPU) other than using hugepage. I'm using IGB UIO module 
with Ethernet 10G 2P X520 Adapter NIC.

I first tried to disable hugepage by inserting "--no-huge" option, but it gave 
an error at rte_mempool_create function. Then I found in rte_eal.h file that 
said "no-huge mode cannot be used with UIO poll-mode drivers like igb/ixgbe".

Is there a way to switch DPDK code to not use hugepage with ixgbe driver or is 
it not possible to do so?

Thank you,
Younghwan

[dpdk-dev] no hugepage with UIO poll-mode driver

[Tan, Jianfeng]

Hi Younghwan
When you specify --no-huge, by default, only 64M memory is used. Can you use -m 
1024 to 
Mmap 1G memory to have a try?

Jianfeng

> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Younghwan Go
> Sent: Wednesday, November 25, 2015 1:46 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] no hugepage with UIO poll-mode driver
> 
> Hello,
> 
> I'm trying to create rte mempool with my own memory scheme (shared
> virtual memory between CPU & GPU) other than using hugepage. I'm using
> IGB UIO module with Ethernet 10G 2P X520 Adapter NIC.
> 
> I first tried to disable hugepage by inserting "--no-huge" option, but it gave
> an error at rte_mempool_create function. Then I found in rte_eal.h file that
> said "no-huge mode cannot be used with UIO poll-mode drivers like
> igb/ixgbe".
> 
> Is there a way to switch DPDK code to not use hugepage with ixgbe driver or
> is it not possible to do so?
> 
> Thank you,
> Younghwan

[dpdk-dev] no hugepage with UIO poll-mode driver

[Younghwan Go]

Hi Jianfeng,

Thanks for the email. rte mempool was successfully created without any 
error. Now the next problem is that rte_eth_rx_burst() is always 
returning 0 as if there was no packet to receive... Do you have any 
suggestion on what might be causing this issue? In the meantime, I will 
be digging through ixgbe driver code to see what's going on.

Thank you,
Younghwan

2015-11-25 ?? 3:19? Tan, Jianfeng ?(?) ? ?:
> Hi Younghwan
> When you specify --no-huge, by default, only 64M memory is used. Can you use 
> -m 1024 to
> Mmap 1G memory to have a try?
>
> Jianfeng
>
>> -Original Message-
>> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Younghwan Go
>> Sent: Wednesday, November 25, 2015 1:46 PM
>> To: dev at dpdk.org
>> Subject: [dpdk-dev] no hugepage with UIO poll-mode driver
>>
>> Hello,
>>
>> I'm trying to create rte mempool with my own memory scheme (shared
>> virtual memory between CPU & GPU) other than using hugepage. I'm using
>> IGB UIO module with Ethernet 10G 2P X520 Adapter NIC.
>>
>> I first tried to disable hugepage by inserting "--no-huge" option, but it 
>> gave
>> an error at rte_mempool_create function. Then I found in rte_eal.h file that
>> said "no-huge mode cannot be used with UIO poll-mode drivers like
>> igb/ixgbe".
>>
>> Is there a way to switch DPDK code to not use hugepage with ixgbe driver or
>> is it not possible to do so?
>>
>> Thank you,
>> Younghwan

[dpdk-dev] [PATCH] mk: fix the combined library problems by replacing it with a linker script

[Panu Matilainen]

On 11/25/2015 12:46 AM, Stephen Hemminger wrote:
> On Tue, 24 Nov 2015 16:31:17 +0200
> Panu Matilainen  wrote:
>
>> The physically linked-together combined library has been an increasing
>> source of problems, as was predicted when library and symbol versioning
>> was introduced. Replace the complex and fragile construction with a
>> simple linker script which achieves the same without all the problems,
>> remove the related kludges from eg mlx drivers.
>>
>> Since creating the linker script is practically zero cost, remove the
>> config option and just create it always.
>>
>> Based on a patch by Sergio Gonzales Monroy, linker script approach
>> initially suggested by Neil Horman.
>>
>> Suggested-by: Sergio Gonzalez Monroy 
>> Suggested-by: Neil Horman 
>> Signed-off-by: Panu Matilainen 
>
> But it now means distros have to ship 20 libraries which seems like
> a step back.

That's how Fedora and RHEL are shipping it already and nobody has so 
much as noticed anything strange, much less complained about it. 20 
libraries is but a drop in the ocean on a average distro. But more to 
the point, distros will prefer 50 working libraries over one that doesn't.

The combined library as it is simply is no longer a viable option. 
Besides just being broken (witness the strange hacks people are coming 
up with to work around issues in it) its ugly because it basically gives 
the middle finger to all the effort going into version compatibility, 
and its also big. Few projects will use every library in DPDK, but with 
the combined library they're forced to lug the 800 pound gorilla along 
needlessly.

- Panu -

[dpdk-dev] Can't compile DPDK if both CONFIG_RTE_BUILD_COMBINE_LIBS and LIBRTE_PMD_XENVIRT are set to "yes"

[Panu Matilainen]

On 11/25/2015 02:26 AM, Martinx - ? wrote:
> On 24 November 2015 at 13:53, Panu Matilainen  wrote:
>> On 11/24/2015 05:30 PM, Martinx - ? wrote:
>>>
>>> On 24 November 2015 at 13:22, Panu Matilainen  
>>> wrote:

 On 11/24/2015 04:46 PM, Sergio Gonzalez Monroy wrote:
>
>
> On 24/11/2015 13:57, Panu Matilainen wrote:
>>
>>
>> On 11/23/2015 08:37 PM, Martinx - ? wrote:
>>>
>>>
>>> Hello!
>>>
>>> My name is Thiago, I'm trying to compile DPDK 2.0, 2.1 and/or 2.2-rc1,
>>> on Ubuntu with Xen support but, it does not build...
>>>
>>> Also, initially, I'm using DPDK sources from Ubuntu APT repository
>>> but, it is also reproducible using upstream DPDK tarball as well,
>>> explained as follows:
>>>
>>> Problem:
>>>
>>> * It is not possible to use the following DPDK options at the same
>>> time:
>>>
>>> CONFIG_RTE_BUILD_COMBINE_LIBS
>>> LIBRTE_PMD_XENVIRT
>>>
>>> Ubuntu DPDK .deb package uses CONFIG_RTE_BUILD_COMBINE_LIBS and,
>>> without it, it can't build its .deb binary package (step: "make -f
>>> debian/rules binary" doesn't work).
>>>
>>> So, if you have the above two options set to "yes", the following
>>> error appear while building DPDK:
>>>
>>> http://pastebin.com/xUsQPxh8
>>>
>> [...]
>>>
>>>
>>> Build error:
>>>
>>> http://pastebin.com/fuUkpF4w
>>>
>>> If you remove "CONFIG_RTE_BUILD_COMBINE_LIBS", then, you can build it
>>> with "LIBRTE_PMD_XENVIRT", and vice-versa. But, without
>>> "...COMBINE_LIBS", Ubuntu .deb package doesn't get builded.
>>>
>>> BTW, the option LIBRTE_XEN_DOM0 is fine when also enabling
>>> COMBINE_LIBS...
>>>
>>> Am I missing something? Is this by design or a DPDK bug?
>>
>>
>>
>> DPDK bug I would say. The combined library has been increasingly in
>> risk of collapsing under its own weight for some time now.
>>
>> A much better way of achieving the same is using a so called linker
>> script which is essentially just an ascii file listing all the
>> individual libraries which the linker handles behind the scenes.
>> FWIW, that's how the combined library is packaged on Fedora and RHEL
>> and consumers like OVS and pktgen never knew the difference.
>>
>> The linker script approach has been suggested before but somehow the
>> threads died without nothing actually happening. I'll revive the patch
>> and post here shortly. Unless Sergio (cc'd) who previously worked on
>> the patches has a newer version cooking silently?
>>
> I haven't worked on it since,  so you probably are in a better position
> to continue the work than me.



 Ok, I suspected as much but thanks for confirming. I'll continue the work
 as
 time permits.

   - Panu -

>>>
>>> Cool! Thank you guys for this fast reply...
>>>
>>> Just a curiosity, do you guys think that this will be ready for 2.2?
>>> Or maybe just for 2.3?
>>>
>>> I'm hoping to use this for next Ubuntu release, Xenial on 2016, April
>>> (Ubuntu 16.04 LTS).
>>>
>>> So, I think that there is time, if not too much trouble for you guys
>>> (honestly, I don't know how hard is to fix that)... :-)
>>
>>
>> Feel free to try this out: http://dpdk.org/dev/patchwork/patch/9088/
>>
>> Whether it lands in time for 2.2 I dont know. However there's absolutely no
>> need to be held hostage by that patch, you can just disable the combined
>> library build option and drop a linker script in its place, created by hand
>> or with couple of lines of shell script, eg:
>> http://pkgs.fedoraproject.org/cgit/dpdk.git/tree/dpdk.spec#n234
>>
>>  - Panu -
>>
>
> Helo Panu,
>
> This is getting interesting... Your patch worked (on top of
> dpdk-2.2.0-rc1) but, now, it broke Ubuntu / Debian packaging, because
> your patch adds a new file called "mk/rte.combinedlib.mk" but it also,
> removes the file "mk/rte.sharelib.mk", which Ubuntu is using to enable
> "LIB ABI Version" on DPDK.
>
> I'm talking about the following Ubuntu Patch (very simple and small):
>
> ---
> tmartins at xenial-1:~/dpdk/ubuntu/dpdk-2.0.0$ cat
> debian/patches/ubuntu-combined-shared-lib-abiversion.patch
>
> http://pastebin.com/VhpMF5U1
> ---
>
> So, with your patch, I can now compile DPDK while using both options
> (_COMBINE_LIBS and _XENVIRT) at the same time but, unfortunately,
> Ubuntu packaging is now broken...
>
> Yes, I know that you guys have nothing to do with that but, I think
> that it would be GREAT if DPDK comes with "Lib ABI Version" by
> default...
>
> So, is it possible to, somehow, include "LIBABIVER" on your new
> "mk/rte.combinedlib.mk" file?
>
> Then, Ubuntu / Debian will not need to patch DPDK anymore, during
> build/packaging process...

The linker script has no ABI version because its not a run-time library, 
its only used during building. The ABI versi

[dpdk-dev] [PATCH] fm10k: add debug info for actual Rx/Tx func

[Chen, Jing D]

> -Original Message-
> From: Stephen Hemminger [mailto:stephen at networkplumber.org]
> Sent: Wednesday, November 25, 2015 12:42 AM
> To: Thomas Monjalon
> Cc: Chen, Jing D; dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH] fm10k: add debug info for actual Rx/Tx func
> 
> On Tue, 24 Nov 2015 11:32:46 +0100
> Thomas Monjalon  wrote:
> 
> > 2015-11-24 14:00, Chen Jing D:
> > > This patch adds debug info to notify user what actual Rx/Tx
> > > func are used.
> > [...]
> > > + if (rx_using_sse)
> > > + PMD_INIT_LOG(ERR, "Use vector Rx func");
> > > + else
> > > + PMD_INIT_LOG(ERR, "Use regular Rx func");
> >
> > debug info != LOG(ERR
> 
> Really should be DEBUG.
> Developers need to remember you don't wan to see those log messages
> in a production system.

Thanks for the comments. I'll change accordingly. 

[dpdk-dev] [PATCH] reserve 'make install' for future use

[Panu Matilainen]

On 11/24/2015 06:54 PM, Bruce Richardson wrote:
> On Fri, Nov 06, 2015 at 02:04:54PM +0100, Thomas Monjalon wrote:
>> 2015-11-06 12:57, Bruce Richardson:
>>> So, any thoughts or comments on this? There has been lots of discussion in 
>>> this
>>> general area but nothing yet going into the release to try and improve the 
>>> situation.
>>>
>>> Are we just going to kick the problem down the road to the 2.3 release?
>>
>> I plan to check these patches in the coming days for an integration in 2.2.
>>
> Anything further on this?
> Any thoughts from anyone else about this whole area of a saner build/install
> system for DPDK and the various patches floating around.

Well, it seems we wont have a sane "make install" in 2.2 yet, but this 
is at least a step in the right direction so +1 from me.

- Panu -

[dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Mcnamara, John]

> -Original Message-
> From: Thomas Monjalon [mailto:thomas.monjalon at 6wind.com]
> Sent: Wednesday, November 25, 2015 1:01 AM
> To: Mcnamara, John
> Cc: dev at dpdk.org; Doherty, Declan
> Subject: Re: [dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new
> QAT DPDK PMD.
> 
> 2015-11-13 18:58, Declan Doherty:
> > +Crypto Device Drivers
> > +
> 
> It is a quite long underlining :)
> 
> It raise a question to me:
> John, have you reviewed the crypto docs?

No. I'll do it now.

John.
-- 

[dpdk-dev] no hugepage with UIO poll-mode driver

[Bruce Richardson]

On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> Hi Jianfeng,
> 
> Thanks for the email. rte mempool was successfully created without any
> error. Now the next problem is that rte_eth_rx_burst() is always returning 0
> as if there was no packet to receive... Do you have any suggestion on what
> might be causing this issue? In the meantime, I will be digging through
> ixgbe driver code to see what's going on.
> 
> Thank you,
> Younghwan
> 

The problem is that with --no-huge we don't have the physical address of the 
memory
to write to the network card. That's what it's marked as for testing only.

/Bruce

> 2015-11-25 ?? 3:19? Tan, Jianfeng ?(?) ? ?:
> >Hi Younghwan
> >When you specify --no-huge, by default, only 64M memory is used. Can you use 
> >-m 1024 to
> >Mmap 1G memory to have a try?
> >
> >Jianfeng
> >
> >>-Original Message-
> >>From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Younghwan Go
> >>Sent: Wednesday, November 25, 2015 1:46 PM
> >>To: dev at dpdk.org
> >>Subject: [dpdk-dev] no hugepage with UIO poll-mode driver
> >>
> >>Hello,
> >>
> >>I'm trying to create rte mempool with my own memory scheme (shared
> >>virtual memory between CPU & GPU) other than using hugepage. I'm using
> >>IGB UIO module with Ethernet 10G 2P X520 Adapter NIC.
> >>
> >>I first tried to disable hugepage by inserting "--no-huge" option, but it 
> >>gave
> >>an error at rte_mempool_create function. Then I found in rte_eal.h file that
> >>said "no-huge mode cannot be used with UIO poll-mode drivers like
> >>igb/ixgbe".
> >>
> >>Is there a way to switch DPDK code to not use hugepage with ixgbe driver or
> >>is it not possible to do so?
> >>
> >>Thank you,
> >>Younghwan
> 

[dpdk-dev] no hugepage with UIO poll-mode driver

[Thomas Monjalon]

2015-11-25 10:08, Bruce Richardson:
> On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> > Hi Jianfeng,
> > 
> > Thanks for the email. rte mempool was successfully created without any
> > error. Now the next problem is that rte_eth_rx_burst() is always returning 0
> > as if there was no packet to receive... Do you have any suggestion on what
> > might be causing this issue? In the meantime, I will be digging through
> > ixgbe driver code to see what's going on.
> > 
> > Thank you,
> > Younghwan
> > 
> 
> The problem is that with --no-huge we don't have the physical address of the 
> memory
> to write to the network card. That's what it's marked as for testing only.

Even with rte_mem_virt2phy() + rte_mem_lock_page() ?

[dpdk-dev] rte_eth_tx_queue_setup() failing (error -22) when setting up tx queues on a VMXNET3 port...

[Montorsi, Francesco]

Hi all,
I have a server running VMWare ESXi 5.5.0 and VMWare vCenter 5.5.0. On such 
server I created a VM with two VMXNET3 NIC cards (one for management, the other 
one should be used with DPDK to enable fast-RX of packets coming from other VMs 
/ bare-metal NICs).

Inside the VM I have successfully created 2MB hugepages, loaded igb_uio driver 
and binded the 2nd VMXNET3 NIC to igb_uio driver:

--
$ ./dpdk_nic_bind.py --status

Network devices using DPDK-compatible driver

:0b:00.0 'VMXNET3 Ethernet Controller' drv=igb_uio unused=vmxnet3

Network devices using kernel driver
===
:03:00.0 'VMXNET3 Ethernet Controller' if=eth0 drv=vmxnet3 unused=igb_uio 
*Active*

Other network devices
=

--

However when I start my DPDK-application I get the following error during TX 
queue initialization:

ERR rte_eth_tx_queue_setup: err=-22, port=0: Unknown error -22

This is the code that I'm using:

  ...
   rte_eth_dev_info_get(m_portid, m_dev_info);


// proceed with configuration

struct rte_eth_conf port_conf;
memset(&port_conf, 0, sizeof(port_conf));

port_conf.rxmode.mq_mode = ETH_MQ_RX_RSS;
/** The multi-queue packet distribution mode to be used, e.g. 
RSS.; this is important to use multiple RX queues per port ID */
port_conf.rxmode.max_rx_pkt_len = ETHER_MAX_LEN;
port_conf.rxmode.hw_strip_crc   = 1; /**< enable CRC stripping by 
hardware */
port_conf.rxmode.jumbo_frame= 1; /**< Jumbo Frame Support enabled */
//port_conf.rxmode.enable_lro = 1; /**< Enable LRO */   
// NOT SUPPORTED ON TESTED HW
port_conf.rx_adv_conf.rss_conf.rss_hf = ETH_RSS_IP;
port_conf.txmode.mq_mode = ETH_MQ_TX_NONE;

m_num_queues = MIN(64, m_dev_info->max_rx_queues);
m_num_queues = MIN(128, m_num_queues);
ret = rte_eth_dev_configure(m_portid, m_num_queues, 1 /* number of tx 
queues */, &port_conf);
if (ret < 0)
{
HMLogError("HwEmulDPDKPort::init() rte_eth_dev_configure: 
err=%d, port=%u: %s", ret, m_portid, rte_strerror(ret));
return false;
}

// init one TX queue: even if we never ever TX packets, at least 1 
queue is needed!
--->ret = rte_eth_tx_queue_setup(m_portid, 0 /* queue ID */, 64 /* num 
descriptors */, rte_eth_dev_socket_id(m_portid), NULL);
if (ret < 0)
{
// retry with just 1 descriptor

--->ret = rte_eth_tx_queue_setup(m_portid, 0 /* queue ID */, 1 /* 
num descriptors */, rte_eth_dev_socket_id(m_portid), NULL);
if (ret < 0)
{
HMLogError("HwEmulDPDKPort::init() 
rte_eth_tx_queue_setup: err=%d, port=%u: %s", ret, m_portid, rte_strerror(ret));
return false;
}
}


Basically since I want to only receive packets (no TX) I'm not really 
interested in TX queues / num of TX descriptors... However I have troubles 
decrypting this -22 error code... any hint?


Thanks a lot,

Francesco Montorsi

[dpdk-dev] [PATCH v7 08/10] aesni_mb_pmd: Initial implementation of multi buffer based crypto device

[Thomas Monjalon]

2015-11-13 18:58, Declan Doherty:
> +To build DPKD with the AESNI_MB_PMD the user is required to download the 
> library
> +from `here `_ and compile 
> it on
> +their user system before building DPDK.

Maybe it is worth saying that yasm must be installed
and compilation is done with "make YASM=yasm" (because of a hard-coded path).

[dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Thomas Monjalon]

2015-11-13 18:58, Declan Doherty:
> +Build and install the SRIOV-enabled QAT driver
> +
> +.. code-block:: console
> +
> +"mkdir /QAT; cd /QAT"
> +copy qatmux.l.2.3.0-34.tgz to this location
> +"tar zxof qatmux.l.2.3.0-34.tgz"
> +"export ICP_WITHOUT_IOMMU=1"
> +"./installer.sh install QAT1.6 host"

People may want to install QAT in a specific directory to just test
build regression.
Is there an easy way to do it?

[dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Thomas Monjalon]

2015-11-25 11:34, Thomas Monjalon:
> 2015-11-13 18:58, Declan Doherty:
> > +Build and install the SRIOV-enabled QAT driver
> > +
> > +.. code-block:: console
> > +
> > +"mkdir /QAT; cd /QAT"
> > +copy qatmux.l.2.3.0-34.tgz to this location
> > +"tar zxof qatmux.l.2.3.0-34.tgz"
> > +"export ICP_WITHOUT_IOMMU=1"
> > +"./installer.sh install QAT1.6 host"
> 
> People may want to install QAT in a specific directory to just test
> build regression.
> Is there an easy way to do it?

For reference, I use this script:

tar xf qatmux-2.5.0-80/QAT1.6/QAT1.6.L.2.5.0-80.tar.gz
export ICP_ROOT=$(readlink -e $qat_dir)
export ICP_ENV_DIR=$ICP_ROOT/quickassist/build_system/build_files/env_files
export ICP_TOOLS_TARGET="accelcomp" 
 
make -C $ICP_ROOT/quickassist

And it fails here:

qat-1.6/quickassist/adf/include/icp_adf_transport_dp.h:118:18:
error: inlining failed in call to always_inline ?icp_adf_pollQueue?:
function body not available

[dpdk-dev] no hugepage with UIO poll-mode driver

[Bruce Richardson]

On Wed, Nov 25, 2015 at 11:23:57AM +0100, Thomas Monjalon wrote:
> 2015-11-25 10:08, Bruce Richardson:
> > On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> > > Hi Jianfeng,
> > > 
> > > Thanks for the email. rte mempool was successfully created without any
> > > error. Now the next problem is that rte_eth_rx_burst() is always 
> > > returning 0
> > > as if there was no packet to receive... Do you have any suggestion on what
> > > might be causing this issue? In the meantime, I will be digging through
> > > ixgbe driver code to see what's going on.
> > > 
> > > Thank you,
> > > Younghwan
> > > 
> > 
> > The problem is that with --no-huge we don't have the physical address of 
> > the memory
> > to write to the network card. That's what it's marked as for testing only.
> 
> Even with rte_mem_virt2phy() + rte_mem_lock_page() ?
> 
With no-huge, we just set up a single memory segment at startup and set its
"physaddr" to be the virtual address.

/Bruce

/* hugetlbfs can be disabled */
if (internal_config.no_hugetlbfs) {
addr = mmap(NULL, internal_config.memory, PROT_READ | 
PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
if (addr == MAP_FAILED) {
RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", __func__,
strerror(errno));
return -1;
}
mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr;
mcfg->memseg[0].addr = addr;
mcfg->memseg[0].hugepage_sz = RTE_PGSIZE_4K;
mcfg->memseg[0].len = internal_config.memory;
mcfg->memseg[0].socket_id = 0;
return 0;
}

[dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Declan Doherty]

On 25/11/15 10:49, Thomas Monjalon wrote:
> 2015-11-25 11:34, Thomas Monjalon:
>> 2015-11-13 18:58, Declan Doherty:
>>> +Build and install the SRIOV-enabled QAT driver
>>> +
>>> +.. code-block:: console
>>> +
>>> +"mkdir /QAT; cd /QAT"
>>> +copy qatmux.l.2.3.0-34.tgz to this location
>>> +"tar zxof qatmux.l.2.3.0-34.tgz"
>>> +"export ICP_WITHOUT_IOMMU=1"
>>> +"./installer.sh install QAT1.6 host"
>>
>> People may want to install QAT in a specific directory to just test
>> build regression.
>> Is there an easy way to do it?
>
> For reference, I use this script:
>
> tar xf qatmux-2.5.0-80/QAT1.6/QAT1.6.L.2.5.0-80.tar.gz
> export ICP_ROOT=$(readlink -e $qat_dir)
> export ICP_ENV_DIR=$ICP_ROOT/quickassist/build_system/build_files/env_files
> export ICP_TOOLS_TARGET="accelcomp"
> make -C $ICP_ROOT/quickassist
>
> And it fails here:
>
> qat-1.6/quickassist/adf/include/icp_adf_transport_dp.h:118:18:
> error: inlining failed in call to always_inline ?icp_adf_pollQueue?:
> function body not available
>


Hey Thomas, I'm just following up with the team on this. There is no 
actual build dependency on the QAT PMD to have the driver installed, 
it's only required for management of the PF and allocation of VF's which 
are subsequently used within DPDK. The only external header dependencies 
outside of DPDK should be on openssl/libcrypto.

Declan

[dpdk-dev] no hugepage with UIO poll-mode driver

[Thomas Monjalon]

2015-11-25 11:00, Bruce Richardson:
> On Wed, Nov 25, 2015 at 11:23:57AM +0100, Thomas Monjalon wrote:
> > 2015-11-25 10:08, Bruce Richardson:
> > > On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> > > > Hi Jianfeng,
> > > > 
> > > > Thanks for the email. rte mempool was successfully created without any
> > > > error. Now the next problem is that rte_eth_rx_burst() is always 
> > > > returning 0
> > > > as if there was no packet to receive... Do you have any suggestion on 
> > > > what
> > > > might be causing this issue? In the meantime, I will be digging through
> > > > ixgbe driver code to see what's going on.
> > > > 
> > > > Thank you,
> > > > Younghwan
> > > > 
> > > 
> > > The problem is that with --no-huge we don't have the physical address of 
> > > the memory
> > > to write to the network card. That's what it's marked as for testing only.
> > 
> > Even with rte_mem_virt2phy() + rte_mem_lock_page() ?
> > 
> With no-huge, we just set up a single memory segment at startup and set its
> "physaddr" to be the virtual address.
> 
> /* hugetlbfs can be disabled */
> if (internal_config.no_hugetlbfs) {
> addr = mmap(NULL, internal_config.memory, PROT_READ | 
> PROT_WRITE,
> MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
> if (addr == MAP_FAILED) {
> RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", __func__,
> strerror(errno));
> return -1;
> }
> mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr;

rte_mem_virt2phy() does not use memseg.phys_addr but /proc/self/pagemap:

/*
 * the pfn (page frame number) are bits 0-54 (see
 * pagemap.txt in linux Documentation)
 */
physaddr = ((page & 0x7fULL) * page_size)
+ ((unsigned long)virtaddr % page_size);

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Ferruh Yigit]

Fixes: 67b6d3039e9e ("eal: set name to threads")

pthread_setname_np() function added in glibc 2.12, using this function
in older glibc versions cause compile error:
error: implicit declaration of function "pthread_setname_np"

This patch adds "rte_thread_setname" macro and set it according
glibc >= 2.12 check, thread naming disabled for older glibc versions,
glibc versions that support "pthread_setname_np" will keep using this
function.

Signed-off-by: Ferruh Yigit 
---
 examples/tep_termination/main.c|  2 +-
 examples/vhost/main.c  |  2 +-
 examples/vhost_xen/main.c  |  2 +-
 lib/librte_eal/common/include/rte_lcore.h  | 20 
 lib/librte_eal/linuxapp/eal/eal.c  |  2 +-
 lib/librte_eal/linuxapp/eal/eal_interrupts.c   |  2 +-
 lib/librte_eal/linuxapp/eal/eal_pci_vfio_mp_sync.c |  2 +-
 lib/librte_eal/linuxapp/eal/eal_timer.c|  2 +-
 8 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/examples/tep_termination/main.c b/examples/tep_termination/main.c
index 2b67e64..f97d552 100644
--- a/examples/tep_termination/main.c
+++ b/examples/tep_termination/main.c
@@ -1249,7 +1249,7 @@ main(int argc, char *argv[])
if (ret != 0)
rte_exit(EXIT_FAILURE, "Cannot create print-stats 
thread\n");
snprintf(thread_name, RTE_MAX_THREAD_NAME_LEN, "print-stats");
-   ret = pthread_setname_np(tid, thread_name);
+   ret = rte_thread_setname(tid, thread_name);
if (ret != 0)
RTE_LOG(ERR, VHOST_CONFIG, "Cannot set print-stats 
name\n");
}
diff --git a/examples/vhost/main.c b/examples/vhost/main.c
index c081b18..9bfda6d 100644
--- a/examples/vhost/main.c
+++ b/examples/vhost/main.c
@@ -3027,7 +3027,7 @@ main(int argc, char *argv[])

/* Set thread_name for aid in debugging.  */
snprintf(thread_name, RTE_MAX_THREAD_NAME_LEN, "print-stats");
-   ret = pthread_setname_np(tid, thread_name);
+   ret = rte_thread_setname(tid, thread_name);
if (ret != 0)
RTE_LOG(ERR, VHOST_CONFIG,
"Cannot set print-stats name\n");
diff --git a/examples/vhost_xen/main.c b/examples/vhost_xen/main.c
index 3fcc138..ca725f2 100644
--- a/examples/vhost_xen/main.c
+++ b/examples/vhost_xen/main.c
@@ -1510,7 +1510,7 @@ main(int argc, char *argv[])

/* Set thread_name for aid in debugging. */
snprintf(thread_name, RTE_MAX_THREAD_NAME_LEN, 
"print-xen-stats");
-   ret = pthread_setname_np(tid, thread_name);
+   ret = rte_thread_setname(tid, thread_name);
if (ret != 0)
RTE_LOG(ERR, VHOST_CONFIG,
"Cannot set print-stats name\n");
diff --git a/lib/librte_eal/common/include/rte_lcore.h 
b/lib/librte_eal/common/include/rte_lcore.h
index e03264e..25460b9 100644
--- a/lib/librte_eal/common/include/rte_lcore.h
+++ b/lib/librte_eal/common/include/rte_lcore.h
@@ -247,6 +247,26 @@ int rte_thread_set_affinity(rte_cpuset_t *cpusetp);
  */
 void rte_thread_get_affinity(rte_cpuset_t *cpusetp);

+/**
+ * Set thread names.
+ *
+ * Macro to wrap `pthread_setname_np()` with a glibc version check.
+ * Only glibc >= 2.12 supports this feature.
+ *
+ * This macro only used for Linux, BSD does direct libc call.
+ * BSD libc version of function is `pthread_set_name_np()`.
+ */
+#if defined(__DOXYGEN__)
+#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
+#endif
+
+#if defined(__GLIBC__) && defined(__GLIBC_PREREQ)
+#if __GLIBC_PREREQ(2, 12)
+#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
+#else
+#define rte_thread_setname(...) 0
+#endif
+#endif

 #ifdef __cplusplus
 }
diff --git a/lib/librte_eal/linuxapp/eal/eal.c 
b/lib/librte_eal/linuxapp/eal/eal.c
index 06536f2..635ec36 100644
--- a/lib/librte_eal/linuxapp/eal/eal.c
+++ b/lib/librte_eal/linuxapp/eal/eal.c
@@ -859,7 +859,7 @@ rte_eal_init(int argc, char **argv)
/* Set thread_name for aid in debugging. */
snprintf(thread_name, RTE_MAX_THREAD_NAME_LEN,
"lcore-slave-%d", i);
-   ret = pthread_setname_np(lcore_config[i].thread_id,
+   ret = rte_thread_setname(lcore_config[i].thread_id,
thread_name);
if (ret != 0)
RTE_LOG(ERR, EAL,
diff --git a/lib/librte_eal/linuxapp/eal/eal_interrupts.c 
b/lib/librte_eal/linuxapp/eal/eal_interrupts.c
index 95beb4c..470d6a1 100644
--- a/lib/librte_eal/linuxapp/eal/eal_interrupts.c
+++ b/lib/librte_eal/linuxapp/eal/eal_interrupts.c
@@ -887,7 +887,7 @@ rte_eal_intr_init(void)
/* Set thread_name for aid in debugging. */
snprintf(thread_name, RTE_MAX_THREAD_NAME_LEN,
 

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Thomas Monjalon]

2015-11-25 11:13, Ferruh Yigit:
> +/**
> + * Set thread names.
> + *
> + * Macro to wrap `pthread_setname_np()` with a glibc version check.
> + * Only glibc >= 2.12 supports this feature.
> + *
> + * This macro only used for Linux, BSD does direct libc call.
> + * BSD libc version of function is `pthread_set_name_np()`.
> + */
> +#if defined(__DOXYGEN__)
> +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> +#endif
> +
> +#if defined(__GLIBC__) && defined(__GLIBC_PREREQ)
> +#if __GLIBC_PREREQ(2, 12)
> +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> +#else
> +#define rte_thread_setname(...) 0
> +#endif
> +#endif

OK it is a first (and important) fix.
EAL is an abstraction for Linux and FreeBSD, so ideally another patch would
make rte_thread_setname working for BSD too.

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Ferruh Yigit]

On Wed, Nov 25, 2015 at 12:18:02PM +0100, Thomas Monjalon wrote:
> 2015-11-25 11:13, Ferruh Yigit:
> > +/**
> > + * Set thread names.
> > + *
> > + * Macro to wrap `pthread_setname_np()` with a glibc version check.
> > + * Only glibc >= 2.12 supports this feature.
> > + *
> > + * This macro only used for Linux, BSD does direct libc call.
> > + * BSD libc version of function is `pthread_set_name_np()`.
> > + */
> > +#if defined(__DOXYGEN__)
> > +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> > +#endif
> > +
> > +#if defined(__GLIBC__) && defined(__GLIBC_PREREQ)
> > +#if __GLIBC_PREREQ(2, 12)
> > +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> > +#else
> > +#define rte_thread_setname(...) 0
> > +#endif
> > +#endif
> 
> OK it is a first (and important) fix.
> EAL is an abstraction for Linux and FreeBSD, so ideally another patch would
> make rte_thread_setname working for BSD too.

I wasn't sure to do that update, since BSD has nothing with glibc versions.
Do you want me amend this patch to update BSD usage to rte_thread_setname?

thanks,
ferruh

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Thomas Monjalon]

2015-11-25 11:24, Ferruh Yigit:
> On Wed, Nov 25, 2015 at 12:18:02PM +0100, Thomas Monjalon wrote:
> > 2015-11-25 11:13, Ferruh Yigit:
> > > +/**
> > > + * Set thread names.
> > > + *
> > > + * Macro to wrap `pthread_setname_np()` with a glibc version check.
> > > + * Only glibc >= 2.12 supports this feature.
> > > + *
> > > + * This macro only used for Linux, BSD does direct libc call.
> > > + * BSD libc version of function is `pthread_set_name_np()`.
> > > + */
> > > +#if defined(__DOXYGEN__)
> > > +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> > > +#endif
> > > +
> > > +#if defined(__GLIBC__) && defined(__GLIBC_PREREQ)
> > > +#if __GLIBC_PREREQ(2, 12)
> > > +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> > > +#else
> > > +#define rte_thread_setname(...) 0
> > > +#endif
> > > +#endif
> > 
> > OK it is a first (and important) fix.
> > EAL is an abstraction for Linux and FreeBSD, so ideally another patch would
> > make rte_thread_setname working for BSD too.
> 
> I wasn't sure to do that update, since BSD has nothing with glibc versions.
> Do you want me amend this patch to update BSD usage to rte_thread_setname?

No, for 2.2, you have fixed the glibc issue, that's enough.
BSD support can be another path as it is not a real issue currently.

[dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Mcnamara, John]

> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Declan Doherty
> Sent: Friday, November 13, 2015 6:58 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v7 07/10] qat_crypto_pmd: Addition of a new QAT
> DPDK PMD.
> 
> This patch adds a PMD for the Intel Quick Assist Technology DH895xxC
> hardware accelerator.




> +Crypto Device Drivers
> +
> +
> +|today|
> +
> +
> +**Contents**
> +

It is best to omit the |today| since it isn't generally useful. Also, 
**Content** is added automatically in the PDF out and not really required in 
Html output so that can be omitted as well.





> +Quick Assist Crypto Poll Mode Driver
> +
> +
> +The QAT PMD provides poll mode crypto driver support for **Intel
> +QuickAssist Technology DH895xxC** hardware accelerator. QAT PMD has
> +current been tested on Fedora 21 64-bit with gcc and on the 4.3
> kernel.org

Typo: current(ly) but it is probably clearer without that word.



> +Features
> +
> +QAT PMD has support for:
> +
> +Cipher algorithms:
> +* RTE_CRYPTO_SYM_CIPHER_AES128_CBC
> +* RTE_CRYPTO_SYM_CIPHER_AES256_CBC
> +* RTE_CRYPTO_SYM_CIPHER_AES512_CBC

The list needs to be separated from the previous line with a blank line to 
render correctly. It would also be worth rending the algorithms as fixed width 
text.

Cipher algorithms:

* ``RTE_CRYPTO_SYM_CIPHER_AES128_CBC``
* ``RTE_CRYPTO_SYM_CIPHER_AES256_CBC``
* ``RTE_CRYPTO_SYM_CIPHER_AES512_CBC``

Same comments for next paragraph.




> +Installation
> +
> +To use the DPDK QAT PMD an SRIOV-enabled QAT kernel driver is required.
> +The VF devices exposed by this driver will be used by QAT PMD.
> +
> +If you are running on kernel 4.3 or greater, see instructions for
> "Installation using
> +kernel.org QAT driver".  If you're on a kernel earlier than 4.3, see
> "Installation using the
> +01.org QAT driver".

The section references don't match the section names. These could also be links 
like this:

If you are running on kernel 4.3 or greater, see instructions for `Installation 
using
kernel.org driver`_ below. If you're on a kernel earlier than 4.3, see 
`Installation using
01.org QAT driver`_.



> +Compiling the 01.org driver - notes:
> +If using a later kernel and the build fails with an error relating to
> strict_stroul not being available patch the following file:
> +
> +.. code-block:: console

You could "use code-block:: diff" here to render the patch nicely in the docs.



> +If build fails due to missing header files you may need to do following:
> +  *  sudo yum install zlib-devel
> +  *  sudo yum install openssl-devel

Probably should be rendered as a code block with ::


> +
> +Installation using kernel.org driver
> +
> +
> +Assuming you are running on at least a 4.3 kernel, you can use the stock
> kernel.org QAT
> +driver to start the QAT hardware.
> +
> +Steps below assume
> +  * running DPDK on a platform with one DH895xCC device
> +  * on a kernel at least version 4.3
> +
> +In BIOS ensure that SRIOV is enabled and VT-d is disabled.
> +
> +Ensure the QAT driver is loaded on your system, by executing:
> +lsmod | grep qat

The commands in this section should be rendered with ::

Ensure the QAT driver is loaded on your system, by executing::

lsmod | grep qat



> +Binding the available VFs to the DPDK UIO driver
> +
> +The unbind command below assumes bdfs of 03:01.00-03:04.07, if yours are
> different adjust the unbind command below.
> +
> +Make available to DPDK
> +
> +.. code-block:: console
> +
> +   cd $(RTE_SDK) (See http://dpdk.org/doc/quick-start to install DPDK)
> +   "modprobe uio"
> +   "insmod ./build/kmod/igb_uio.ko"
> +   "for device in $(seq 1 4); do for fn in $(seq 0 7); do echo -n
> :03:0${device}.${fn} >
> /sys/bus/pci/devices/\:03\:0${device}.${fn}/driver/unbind;done ;done"
> +   "echo "8086 0443" > /sys/bus/pci/drivers/igb_uio/new_id"
> +

This is too long to be rendered in PDF. Something like the following would work 
better in the docs while being functionally the same:

The unbind command below assumes ``bdfs`` of ``03:01.00-03:04.07``, if yours 
are different adjust the unbind command below::

   cd $RTE_SDK
   modprobe uio
   insmod ./build/kmod/igb_uio.ko

   for device in $(seq 1 4); do \
   for fn in $(seq 0 7); do \
   echo -n :03:0${device}.${fn} > \
   /sys/bus/pci/devices/\:03\:0${device}.${fn}/driver/unbind; \
   done; \
   done

   echo "8086 0443" > /sys/bus/pci/drivers/igb_uio/new_id

You can use ``lspci -vvd:443`` to confirm that all devices are now in use by 
igb_uio kernel driver.


John.
-- 

[dpdk-dev] no hugepage with UIO poll-mode driver

[Bruce Richardson]

On Wed, Nov 25, 2015 at 12:03:05PM +0100, Thomas Monjalon wrote:
> 2015-11-25 11:00, Bruce Richardson:
> > On Wed, Nov 25, 2015 at 11:23:57AM +0100, Thomas Monjalon wrote:
> > > 2015-11-25 10:08, Bruce Richardson:
> > > > On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> > > > > Hi Jianfeng,
> > > > > 
> > > > > Thanks for the email. rte mempool was successfully created without any
> > > > > error. Now the next problem is that rte_eth_rx_burst() is always 
> > > > > returning 0
> > > > > as if there was no packet to receive... Do you have any suggestion on 
> > > > > what
> > > > > might be causing this issue? In the meantime, I will be digging 
> > > > > through
> > > > > ixgbe driver code to see what's going on.
> > > > > 
> > > > > Thank you,
> > > > > Younghwan
> > > > > 
> > > > 
> > > > The problem is that with --no-huge we don't have the physical address 
> > > > of the memory
> > > > to write to the network card. That's what it's marked as for testing 
> > > > only.
> > > 
> > > Even with rte_mem_virt2phy() + rte_mem_lock_page() ?
> > > 
> > With no-huge, we just set up a single memory segment at startup and set its
> > "physaddr" to be the virtual address.
> > 
> > /* hugetlbfs can be disabled */
> > if (internal_config.no_hugetlbfs) {
> > addr = mmap(NULL, internal_config.memory, PROT_READ | 
> > PROT_WRITE,
> > MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
> > if (addr == MAP_FAILED) {
> > RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", 
> > __func__,
> > strerror(errno));
> > return -1;
> > }
> > mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr;
> 
> rte_mem_virt2phy() does not use memseg.phys_addr but /proc/self/pagemap:
> 
> /*
>  * the pfn (page frame number) are bits 0-54 (see
>  * pagemap.txt in linux Documentation)
>  */
> physaddr = ((page & 0x7fULL) * page_size)
> + ((unsigned long)virtaddr % page_size);
>

Yes, you are right. I was not aware that that function was used as part of the
mempool init, but now I see that "rte_mempool_virt2phy()" does indeed call that
function if hugepages are disabled, so my bad.

/Bruce

[dpdk-dev] [PATCH] mk: fix the combined library problems by replacing it with a linker script

[Neil Horman]

On Wed, Nov 25, 2015 at 10:38:48AM +0200, Panu Matilainen wrote:
> On 11/25/2015 12:46 AM, Stephen Hemminger wrote:
> >On Tue, 24 Nov 2015 16:31:17 +0200
> >Panu Matilainen  wrote:
> >
> >>The physically linked-together combined library has been an increasing
> >>source of problems, as was predicted when library and symbol versioning
> >>was introduced. Replace the complex and fragile construction with a
> >>simple linker script which achieves the same without all the problems,
> >>remove the related kludges from eg mlx drivers.
> >>
> >>Since creating the linker script is practically zero cost, remove the
> >>config option and just create it always.
> >>
> >>Based on a patch by Sergio Gonzales Monroy, linker script approach
> >>initially suggested by Neil Horman.
> >>
> >>Suggested-by: Sergio Gonzalez Monroy 
> >>Suggested-by: Neil Horman 
> >>Signed-off-by: Panu Matilainen 
> >
> >But it now means distros have to ship 20 libraries which seems like
> >a step back.
> 
> That's how Fedora and RHEL are shipping it already and nobody has so much as
> noticed anything strange, much less complained about it. 20 libraries is but
> a drop in the ocean on a average distro. But more to the point, distros will
> prefer 50 working libraries over one that doesn't.
> 
> The combined library as it is simply is no longer a viable option. Besides
> just being broken (witness the strange hacks people are coming up with to
> work around issues in it) its ugly because it basically gives the middle
> finger to all the effort going into version compatibility, and its also big.
> Few projects will use every library in DPDK, but with the combined library
> they're forced to lug the 800 pound gorilla along needlessly.
> 
Agreed,  This solves a ton of problems, and from a distro standpoint, no one
really cares how many libraries it is under the covers.  Its all just one
rpm/deb package anyway.

Neil

[dpdk-dev] no hugepage with UIO poll-mode driver

[Thomas Monjalon]

2015-11-25 12:02, Bruce Richardson:
> On Wed, Nov 25, 2015 at 12:03:05PM +0100, Thomas Monjalon wrote:
> > 2015-11-25 11:00, Bruce Richardson:
> > > On Wed, Nov 25, 2015 at 11:23:57AM +0100, Thomas Monjalon wrote:
> > > > 2015-11-25 10:08, Bruce Richardson:
> > > > > On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> > > > > > Hi Jianfeng,
> > > > > > 
> > > > > > Thanks for the email. rte mempool was successfully created without 
> > > > > > any
> > > > > > error. Now the next problem is that rte_eth_rx_burst() is always 
> > > > > > returning 0
> > > > > > as if there was no packet to receive... Do you have any suggestion 
> > > > > > on what
> > > > > > might be causing this issue? In the meantime, I will be digging 
> > > > > > through
> > > > > > ixgbe driver code to see what's going on.
> > > > > > 
> > > > > > Thank you,
> > > > > > Younghwan
> > > > > > 
> > > > > 
> > > > > The problem is that with --no-huge we don't have the physical address 
> > > > > of the memory
> > > > > to write to the network card. That's what it's marked as for testing 
> > > > > only.
> > > > 
> > > > Even with rte_mem_virt2phy() + rte_mem_lock_page() ?
> > > > 
> > > With no-huge, we just set up a single memory segment at startup and set 
> > > its
> > > "physaddr" to be the virtual address.
> > > 
> > > /* hugetlbfs can be disabled */
> > > if (internal_config.no_hugetlbfs) {
> > > addr = mmap(NULL, internal_config.memory, PROT_READ | 
> > > PROT_WRITE,
> > > MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
> > > if (addr == MAP_FAILED) {
> > > RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", 
> > > __func__,
> > > strerror(errno));
> > > return -1;
> > > }
> > > mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr;
> > 
> > rte_mem_virt2phy() does not use memseg.phys_addr but /proc/self/pagemap:
> > 
> > /*
> >  * the pfn (page frame number) are bits 0-54 (see
> >  * pagemap.txt in linux Documentation)
> >  */
> > physaddr = ((page & 0x7fULL) * page_size)
> > + ((unsigned long)virtaddr % page_size);
> >
> 
> Yes, you are right. I was not aware that that function was used as part of the
> mempool init, but now I see that "rte_mempool_virt2phy()" does indeed call 
> that
> function if hugepages are disabled, so my bad.

Do you think we could move --no-huge in the main section (not only for testing)?

[dpdk-dev] [PATCH v8 00/10] Crypto API and device framework

[Declan Doherty]

This series of patches defines a set of application burst oriented APIs for
asynchronous symmetric cryptographic functions within DPDK. It also contains a
poll mode driver cryptographic device framework for the implementation of
crypto devices within DPDK.

In the patch set we also have included 2 reference implementations of crypto
PMDs. Currently both implementations support AES-CBC with
HMAC_SHA1/SHA256/SHA512 authentication operations. The first device is a purely
software PMD based on Intel's multi-buffer library, which utilises both
AES-NI instructions and vector operations to accelerate crypto operations and
the second PMD utilises Intel's Quick Assist Technology (on DH895xxC) to
provide hardware accelerated crypto operations.

The API set supports two functional modes of operation:

1, A session oriented mode. In this mode the user creates a crypto session
which defines all the immutable data required to perform a particular crypto
operation in advance, including cipher/hash algorithms and operations to be
performed as well as the keys to used etc. The session is then referenced by
the crypto operation data structure which is a data structure specific to each
mbuf. It is contains all mutable data about the crypto operation to be
performed, such as data offsets and lengths into the mbuf's data payload for
cipher and hash operations to be performed.

2, A session-less mode. In this mode the user is able to provision crypto
operations on an mbuf without the need to have a cached session created in
advance, but at the cost of entailing the overhead of calculating
authentication pre-computes and preforming key expansions in-line with the
crypto operation. The crypto xform chain is directly attached to the op struct
in this mode, so the op struct now contains all of the immutable crypto
operation parameters that would be normally set within a session. Once all
mutable and immutable parameters are set the crypto operation data structure
can be attached to the specified mbuf and enqueued on a specified crypto device
for processing.

The patch set contains the following features:
 - Crypto device APIs and device framework
 - Implementation of a software crypto PMD based on multi-buffer library
 - Implementation of a hardware crypto PMD baed on Intel QAT(DH895xxC)
 - Unit and performance test's which give and example of utilising the crypto
 API's.
 - Sample application which performs crypto operations on the IP payload of the
   packets being forwarded

Current Status:
There is no support for chained mbuf's and as mentioned above the PMD's
have currently implemented support for AES128-CBC/AES192-CBC/AES256-CBC
and HMAC_SHA1/SHA256/SHA512 and AES_XCBC_MAC.

v8:
  - Doxygen comment fix for rte_pktmbuf_mtophys macro
  - Doxygen fixes for public headers in rte_crypto.h
  - QAT documentation tidy up based on J. McNamara comments
  - Detailed requirement to set YASM path when building multi-buffer  library
  - l2fwd-crypto: fix for 32-bit build; fix for possible memory leak if 
rte_cryptodev_burst_enqueue fails; and handling for failure to  allocate
rte_mbuf_offload.

v7:
  - Fix typos in commit message of eal: add __rte_packed /__rte_aligned macros 
patch
  - Include rte_mbuf_offload in doxygen build and updates file comments to 
clarify lib,
usage. Also moved clean which was in wrong patch into this
rte_mbuf_offload patch.
  - Tidy up map file for cryptodev library.
  - Add l2fwdc-crypto to main examples makefile.
v6:
  - Fix 32-bit build issue caused by casting in new  rte_pktmbuf_mtophys_offset 
macro
  - Fix truncation of log message by new  rte_pmd_debug_trace inline function

v5:
  - Making ethdev marcos for function pointer and port id checking public and
available for use in by the cryptodev. The intialise to patches combine 
changes
from original cryptodev patch and discussion in
http://dpdk.org/ml/archives/dev/2015-November/027871.html
  - Split out changes to create new __rte_packed and __rte_aligned macros 
into seperate patches form the main  cryptodev patch set for clairty
  - further code cleaning, removal of currently unsupported gcm code from  
aesni_mb pmd

v4:
  - Some more EOF whitespace and  checkpatch fixes

v3:
  - Fixes a document build error, which I missed in the V2
  - Fixes for remaining checkpatch errors
  - Disables QAT and AESNI_MB PMD being build by default as they have external 
library dependences 

v2: 
 - Introduces a new library to support attaching offload operations to a mbuf
 - Remove unused APIs from cryptodev
 - PMD code refactor due to  new rte_mbuf_offload structure
 - General bug fixes and code tidy up


Declan Doherty (10):
  ethdev: rename macros to have RTE_ prefix
  ethdev: make error checking macros public
  eal: add __rte_packed /__rte_aligned macros
  mbuf: add new marcos to get the physical address of data
  cryptodev: Initial DPDK Crypto APIs and device framework release
  mbuf_offload: library to support attaching offloads

[dpdk-dev] [PATCH v8 01/10] ethdev: rename macros to have RTE_ prefix

[Declan Doherty]

The macros to check that the function pointers and port ids are valid
for an ethdev are potentially useful to have in a common headers for
use with all PMDs. However, since they would then become externally
visible, we apply the RTE_ & RTE_ETH_ prefix to them as approtiate.

Signed-off-by: Declan Doherty 
Acked-by: Bruce Richardson 

---
 lib/librte_ether/rte_ethdev.c | 607 +-
 1 file changed, 304 insertions(+), 303 deletions(-)

diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index b19ac9a..71775dc 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -70,58 +70,59 @@
 #include "rte_ethdev.h"

 #ifdef RTE_LIBRTE_ETHDEV_DEBUG
-#define PMD_DEBUG_TRACE(fmt, args...) do {\
+#define RTE_PMD_DEBUG_TRACE(fmt, args...) do { \
RTE_LOG(ERR, PMD, "%s: " fmt, __func__, ## args); \
} while (0)
 #else
-#define PMD_DEBUG_TRACE(fmt, args...)
+#define RTE_PMD_DEBUG_TRACE(fmt, args...)
 #endif

 /* Macros for checking for restricting functions to primary instance only */
-#define PROC_PRIMARY_OR_ERR_RET(retval) do { \
+#define RTE_PROC_PRIMARY_OR_ERR_RET(retval) do { \
if (rte_eal_process_type() != RTE_PROC_PRIMARY) { \
-   PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
+   RTE_PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
return (retval); \
} \
 } while (0)

-#define PROC_PRIMARY_OR_RET() do { \
+#define RTE_PROC_PRIMARY_OR_RET() do { \
if (rte_eal_process_type() != RTE_PROC_PRIMARY) { \
-   PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
+   RTE_PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
return; \
} \
 } while (0)

 /* Macros to check for invalid function pointers in dev_ops structure */
-#define FUNC_PTR_OR_ERR_RET(func, retval) do { \
+#define RTE_FUNC_PTR_OR_ERR_RET(func, retval) do { \
if ((func) == NULL) { \
-   PMD_DEBUG_TRACE("Function not supported\n"); \
+   RTE_PMD_DEBUG_TRACE("Function not supported\n"); \
return (retval); \
} \
 } while (0)

-#define FUNC_PTR_OR_RET(func) do { \
+#define RTE_FUNC_PTR_OR_RET(func) do { \
if ((func) == NULL) { \
-   PMD_DEBUG_TRACE("Function not supported\n"); \
+   RTE_PMD_DEBUG_TRACE("Function not supported\n"); \
return; \
} \
 } while (0)

 /* Macros to check for valid port */
-#define VALID_PORTID_OR_ERR_RET(port_id, retval) do {  \
-   if (!rte_eth_dev_is_valid_port(port_id)) {  \
-   PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
-   return retval;  \
-   }   \
+#define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
+   if (!rte_eth_dev_is_valid_port(port_id)) {  \
+   RTE_PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
+   return retval; \
+   } \
 } while (0)

-#define VALID_PORTID_OR_RET(port_id) do {  \
-   if (!rte_eth_dev_is_valid_port(port_id)) {  \
-   PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
-   return; \
-   }   \
+#define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
+   if (!rte_eth_dev_is_valid_port(port_id)) { \
+   RTE_PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
+   return; \
+   } \
 } while (0)

+
 static const char *MZ_RTE_ETH_DEV_DATA = "rte_eth_dev_data";
 struct rte_eth_dev rte_eth_devices[RTE_MAX_ETHPORTS];
 static struct rte_eth_dev_data *rte_eth_dev_data;
@@ -244,7 +245,7 @@ rte_eth_dev_allocate(const char *name, enum 
rte_eth_dev_type type)

port_id = rte_eth_dev_find_free_port();
if (port_id == RTE_MAX_ETHPORTS) {
-   PMD_DEBUG_TRACE("Reached maximum number of Ethernet ports\n");
+   RTE_PMD_DEBUG_TRACE("Reached maximum number of Ethernet 
ports\n");
return NULL;
}

@@ -252,7 +253,7 @@ rte_eth_dev_allocate(const char *name, enum 
rte_eth_dev_type type)
rte_eth_dev_data_alloc();

if (rte_eth_dev_allocated(name) != NULL) {
-   PMD_DEBUG_TRACE("Ethernet Device with name %s already 
allocated!\n",
+   RTE_PMD_DEBUG_TRACE("Ethernet Device with name %s already 
allocated!\n",
name);
return NULL;
}
@@ -339,7 +340,7 @@ rte_eth_dev_init(struct rte_pci_driver *pci_drv,
if (diag == 0)
return 0;

-   PMD_DEBUG_TRACE("driver %s: eth_dev_init(vendor_id=0x%u device_id=0x%x) 
failed\n",
+   RTE_PMD_DEBUG_TRACE("driver %s: eth_dev_init(vendor_id=0x%u 
device_id=0x%x) failed\n

[dpdk-dev] [PATCH v8 03/10] eal: add __rte_packed /__rte_aligned macros

[Declan Doherty]

Adding a new macro for specifying __aligned__ attribute, and updating the
current __rte_cache_aligned macro to use it.

Also adding a new macro to specify the __packed__ attribute

Signed-off-by: Declan Doherty 
Acked-by: Sergio Gonzalez Monroy 

---
 lib/librte_eal/common/include/rte_memory.h | 14 --
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/lib/librte_eal/common/include/rte_memory.h 
b/lib/librte_eal/common/include/rte_memory.h
index 067be10..20feed9 100644
--- a/lib/librte_eal/common/include/rte_memory.h
+++ b/lib/librte_eal/common/include/rte_memory.h
@@ -78,9 +78,19 @@ enum rte_page_sizes {
 /**< Return the first cache-aligned value greater or equal to size. */

 /**
+ * Force alignment
+ */
+#define __rte_aligned(a) __attribute__((__aligned__(a)))
+
+/**
  * Force alignment to cache line.
  */
-#define __rte_cache_aligned __attribute__((__aligned__(RTE_CACHE_LINE_SIZE)))
+#define __rte_cache_aligned __rte_aligned(RTE_CACHE_LINE_SIZE)
+
+/**
+ * Force a structure to be packed
+ */
+#define __rte_packed __attribute__((__packed__))

 typedef uint64_t phys_addr_t; /**< Physical address definition. */
 #define RTE_BAD_PHYS_ADDR ((phys_addr_t)-1)
@@ -106,7 +116,7 @@ struct rte_memseg {
 /**< store segment MFNs */
uint64_t mfn[DOM0_NUM_MEMBLOCK];
 #endif
-} __attribute__((__packed__));
+} __rte_packed;

 /**
  * Lock page in physical memory and prevent from swapping.
-- 
2.5.0

[dpdk-dev] [PATCH v8 04/10] mbuf: add new marcos to get the physical address of data

[Declan Doherty]

Signed-off-by: Declan Doherty 
Acked-by: Sergio Gonzalez Monroy 

---
 lib/librte_mbuf/rte_mbuf.h | 21 +
 1 file changed, 21 insertions(+)

diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index 4a93189..6a1c133 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -1622,6 +1622,27 @@ static inline struct rte_mbuf 
*rte_pktmbuf_lastseg(struct rte_mbuf *m)
 #define rte_pktmbuf_mtod(m, t) rte_pktmbuf_mtod_offset(m, t, 0)

 /**
+ * A macro that returns the physical address that points to an offset of the
+ * start of the data in the mbuf
+ *
+ * @param m
+ *   The packet mbuf.
+ * @param o
+ *   The offset into the data to calculate address from.
+ */
+#define rte_pktmbuf_mtophys_offset(m, o) \
+   (phys_addr_t)((m)->buf_physaddr + (m)->data_off + (o))
+
+/**
+ * A macro that returns the physical address that points to the start of the
+ * data in the mbuf
+ *
+ * @param m
+ *   The packet mbuf.
+ */
+#define rte_pktmbuf_mtophys(m) rte_pktmbuf_mtophys_offset(m, 0)
+
+/**
  * A macro that returns the length of the packet.
  *
  * The value can be read or assigned.
-- 
2.5.0

[dpdk-dev] [PATCH v8 02/10] ethdev: make error checking macros public

[Declan Doherty]

Move the function pointer and port id checking macros to rte_ethdev and
rte_dev header files, so that they can be used in the static inline
functions there. Also replace the RTE_LOG call within
RTE_PMD_DEBUG_TRACE so this macro can be built with the -pedantic flag

Signed-off-by: Declan Doherty 
Acked-by: Adrien Mazarguil 

---
 lib/librte_eal/common/include/rte_dev.h | 53 
 lib/librte_ether/rte_ethdev.c   | 54 -
 lib/librte_ether/rte_ethdev.h   | 26 
 3 files changed, 79 insertions(+), 54 deletions(-)

diff --git a/lib/librte_eal/common/include/rte_dev.h 
b/lib/librte_eal/common/include/rte_dev.h
index f601d21..f1b5507 100644
--- a/lib/librte_eal/common/include/rte_dev.h
+++ b/lib/librte_eal/common/include/rte_dev.h
@@ -46,8 +46,61 @@
 extern "C" {
 #endif

+#include 
 #include 

+#include 
+
+__attribute__((format(printf, 2, 0)))
+static inline void
+rte_pmd_debug_trace(const char *func_name, const char *fmt, ...)
+{
+   va_list ap;
+
+   va_start(ap, fmt);
+
+   char buffer[vsnprintf(NULL, 0, fmt, ap) + 1];
+
+   va_end(ap);
+
+   va_start(ap, fmt);
+   vsnprintf(buffer, sizeof(buffer), fmt, ap);
+   va_end(ap);
+
+   rte_log(RTE_LOG_ERR, RTE_LOGTYPE_PMD, "%s: %s", func_name, buffer);
+}
+
+/* Macros for checking for restricting functions to primary instance only */
+#define RTE_PROC_PRIMARY_OR_ERR_RET(retval) do { \
+   if (rte_eal_process_type() != RTE_PROC_PRIMARY) { \
+   RTE_PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
+   return retval; \
+   } \
+} while (0)
+
+#define RTE_PROC_PRIMARY_OR_RET() do { \
+   if (rte_eal_process_type() != RTE_PROC_PRIMARY) { \
+   RTE_PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
+   return; \
+   } \
+} while (0)
+
+/* Macros to check for invalid function pointers */
+#define RTE_FUNC_PTR_OR_ERR_RET(func, retval) do { \
+   if ((func) == NULL) { \
+   RTE_PMD_DEBUG_TRACE("Function not supported\n"); \
+   return retval; \
+   } \
+} while (0)
+
+#define RTE_FUNC_PTR_OR_RET(func) do { \
+   if ((func) == NULL) { \
+   RTE_PMD_DEBUG_TRACE("Function not supported\n"); \
+   return; \
+   } \
+} while (0)
+
+
 /** Double linked list of device drivers. */
 TAILQ_HEAD(rte_driver_list, rte_driver);

diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index 71775dc..f4648ac 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -69,60 +69,6 @@
 #include "rte_ether.h"
 #include "rte_ethdev.h"

-#ifdef RTE_LIBRTE_ETHDEV_DEBUG
-#define RTE_PMD_DEBUG_TRACE(fmt, args...) do { \
-   RTE_LOG(ERR, PMD, "%s: " fmt, __func__, ## args); \
-   } while (0)
-#else
-#define RTE_PMD_DEBUG_TRACE(fmt, args...)
-#endif
-
-/* Macros for checking for restricting functions to primary instance only */
-#define RTE_PROC_PRIMARY_OR_ERR_RET(retval) do { \
-   if (rte_eal_process_type() != RTE_PROC_PRIMARY) { \
-   RTE_PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
-   return (retval); \
-   } \
-} while (0)
-
-#define RTE_PROC_PRIMARY_OR_RET() do { \
-   if (rte_eal_process_type() != RTE_PROC_PRIMARY) { \
-   RTE_PMD_DEBUG_TRACE("Cannot run in secondary processes\n"); \
-   return; \
-   } \
-} while (0)
-
-/* Macros to check for invalid function pointers in dev_ops structure */
-#define RTE_FUNC_PTR_OR_ERR_RET(func, retval) do { \
-   if ((func) == NULL) { \
-   RTE_PMD_DEBUG_TRACE("Function not supported\n"); \
-   return (retval); \
-   } \
-} while (0)
-
-#define RTE_FUNC_PTR_OR_RET(func) do { \
-   if ((func) == NULL) { \
-   RTE_PMD_DEBUG_TRACE("Function not supported\n"); \
-   return; \
-   } \
-} while (0)
-
-/* Macros to check for valid port */
-#define RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, retval) do { \
-   if (!rte_eth_dev_is_valid_port(port_id)) {  \
-   RTE_PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
-   return retval; \
-   } \
-} while (0)
-
-#define RTE_ETH_VALID_PORTID_OR_RET(port_id) do { \
-   if (!rte_eth_dev_is_valid_port(port_id)) { \
-   RTE_PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id); \
-   return; \
-   } \
-} while (0)
-
-
 static const char *MZ_RTE_ETH_DEV_DATA = "rte_eth_dev_data";
 struct rte_eth_dev rte_eth_devices[RTE_MAX_ETHPORTS];
 static struct rte_eth_dev_data *rte_eth_dev_data;
diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index e92bf8d..b51b8aa 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -172,6 +172,8 @@ extern "C" {

 #include 

+#include 
+
 /* Use this macro to check if LRO API is supported */
 #define RTE_ETHDEV_HAS_LRO_SUPPORT

@@ -931

[dpdk-dev] [PATCH v8 06/10] mbuf_offload: library to support attaching offloads to a mbuf

[Declan Doherty]

This library add support for adding a chain of offload operations to a
mbuf. It contains the definition of the rte_mbuf_offload structure as
well as helper functions for attaching  offloads to mbufs and a mempool
management functions.

This initial implementation supports attaching multiple offload
operations to a single mbuf, but only a single offload operation of a
specific type can be attach to that mbuf.

Signed-off-by: Declan Doherty 
Acked-by: Konstantin Ananyev 

---
 MAINTAINERS|   4 +
 config/common_bsdapp   |   6 +
 config/common_linuxapp |   6 +
 doc/api/doxy-api-index.md  |   1 +
 doc/api/doxy-api.conf  |   1 +
 lib/Makefile   |   1 +
 lib/librte_mbuf/rte_mbuf.h |   6 +
 lib/librte_mbuf_offload/Makefile   |  52 
 lib/librte_mbuf_offload/rte_mbuf_offload.c | 100 +++
 lib/librte_mbuf_offload/rte_mbuf_offload.h | 302 +
 .../rte_mbuf_offload_version.map   |   7 +
 mk/rte.app.mk  |   1 +
 12 files changed, 487 insertions(+)
 create mode 100644 lib/librte_mbuf_offload/Makefile
 create mode 100644 lib/librte_mbuf_offload/rte_mbuf_offload.c
 create mode 100644 lib/librte_mbuf_offload/rte_mbuf_offload.h
 create mode 100644 lib/librte_mbuf_offload/rte_mbuf_offload_version.map

diff --git a/MAINTAINERS b/MAINTAINERS
index 9138bbb..dd8be0f 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -191,6 +191,10 @@ F: lib/librte_mbuf/
 F: doc/guides/prog_guide/mbuf_lib.rst
 F: app/test/test_mbuf.c

+Packet buffer offload
+M: Declan Doherty 
+F: lib/librte_mbuf_offload/
+
 Ethernet API
 M: Thomas Monjalon 
 F: lib/librte_ether/
diff --git a/config/common_bsdapp b/config/common_bsdapp
index 3bfb3f6..e536fdf 100644
--- a/config/common_bsdapp
+++ b/config/common_bsdapp
@@ -335,6 +335,12 @@ CONFIG_RTE_MBUF_REFCNT_ATOMIC=y
 CONFIG_RTE_PKTMBUF_HEADROOM=128

 #
+# Compile librte_mbuf_offload
+#
+CONFIG_RTE_LIBRTE_MBUF_OFFLOAD=y
+CONFIG_RTE_LIBRTE_MBUF_OFFLOAD_DEBUG=n
+
+#
 # Compile librte_timer
 #
 CONFIG_RTE_LIBRTE_TIMER=y
diff --git a/config/common_linuxapp b/config/common_linuxapp
index cd7a2d4..1947ce3 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -343,6 +343,12 @@ CONFIG_RTE_MBUF_REFCNT_ATOMIC=y
 CONFIG_RTE_PKTMBUF_HEADROOM=128

 #
+# Compile librte_mbuf_offload
+#
+CONFIG_RTE_LIBRTE_MBUF_OFFLOAD=y
+CONFIG_RTE_LIBRTE_MBUF_OFFLOAD_DEBUG=n
+
+#
 # Compile librte_timer
 #
 CONFIG_RTE_LIBRTE_TIMER=y
diff --git a/doc/api/doxy-api-index.md b/doc/api/doxy-api-index.md
index bdb6130..199cc2c 100644
--- a/doc/api/doxy-api-index.md
+++ b/doc/api/doxy-api-index.md
@@ -104,6 +104,7 @@ There are many libraries, so their headers may be grouped 
by topics:

 - **containers**:
   [mbuf]   (@ref rte_mbuf.h),
+  [mbuf_offload]   (@ref rte_mbuf_offload.h),
   [ring]   (@ref rte_ring.h),
   [distributor](@ref rte_distributor.h),
   [reorder](@ref rte_reorder.h),
diff --git a/doc/api/doxy-api.conf b/doc/api/doxy-api.conf
index 7244b8f..15bba16 100644
--- a/doc/api/doxy-api.conf
+++ b/doc/api/doxy-api.conf
@@ -48,6 +48,7 @@ INPUT   = doc/api/doxy-api-index.md \
   lib/librte_kvargs \
   lib/librte_lpm \
   lib/librte_mbuf \
+  lib/librte_mbuf_offload \
   lib/librte_mempool \
   lib/librte_meter \
   lib/librte_net \
diff --git a/lib/Makefile b/lib/Makefile
index 4c5c1b4..ef172ea 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -36,6 +36,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_EAL) += librte_eal
 DIRS-$(CONFIG_RTE_LIBRTE_RING) += librte_ring
 DIRS-$(CONFIG_RTE_LIBRTE_MEMPOOL) += librte_mempool
 DIRS-$(CONFIG_RTE_LIBRTE_MBUF) += librte_mbuf
+DIRS-$(CONFIG_RTE_LIBRTE_MBUF_OFFLOAD) += librte_mbuf_offload
 DIRS-$(CONFIG_RTE_LIBRTE_TIMER) += librte_timer
 DIRS-$(CONFIG_RTE_LIBRTE_CFGFILE) += librte_cfgfile
 DIRS-$(CONFIG_RTE_LIBRTE_CMDLINE) += librte_cmdline
diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index 6a1c133..cb4583d 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -728,6 +728,9 @@ typedef uint8_t  MARKER8[0];  /**< generic marker with 1B 
alignment */
 typedef uint64_t MARKER64[0]; /**< marker that allows us to overwrite 8 bytes
* with a single assignment */

+/** Opaque rte_mbuf_offload  structure declarations */
+struct rte_mbuf_offload;
+
 /**
  * The generic rte_mbuf, containing a packet mbuf.
  */
@@ -841,6 +844,9 @@ struct rte_mbuf {

/** Timesync flags for use with IEEE1588. */
uint16_t timesync;
+
+   /* Chain of off-load operations to perform on mbuf */
+   struct rte_mbuf

[dpdk-dev] [PATCH v8 05/10] cryptodev: Initial DPDK Crypto APIs and device framework release

[Declan Doherty]

This patch contains the initial proposed APIs and device framework for
integrating crypto packet processing into DPDK.

features include:
 - Crypto device configuration / management APIs
 - Definitions of supported cipher algorithms and operations.
 - Definitions of supported hash/authentication algorithms and
   operations.
 - Crypto session management APIs
 - Crypto operation data structures and APIs allocation of crypto
   operation structure used to specify the crypto operations to
   be performed  on a particular mbuf.
 - Extension of mbuf to contain crypto operation data pointer and
   extra flags.
 - Burst enqueue / dequeue APIs for processing of crypto operations.

changes from RFC:
 - Session management API changes to support specification of crypto
   transform(xform) chains using linked list of xforms.
 - Changes to the crypto operation struct as a result of session
   management changes.
 - Some movement of common MACROS shared by cryptodevs and ethdevs to
   common headers

Signed-off-by: Des O Dea 
Signed-off-by: John Griffin 
Signed-off-by: Fiona Trahe 
Signed-off-by: Declan Doherty 

Acked-by: Sergio Gonzalez Monroy 

---
 MAINTAINERS|4 +
 config/common_bsdapp   |   10 +-
 config/common_linuxapp |   10 +-
 doc/api/doxy-api-index.md  |1 +
 doc/api/doxy-api.conf  |1 +
 lib/Makefile   |1 +
 lib/librte_cryptodev/Makefile  |   60 ++
 lib/librte_cryptodev/rte_crypto.h  |  610 +
 lib/librte_cryptodev/rte_cryptodev.c   | 1092 
 lib/librte_cryptodev/rte_cryptodev.h   |  651 ++
 lib/librte_cryptodev/rte_cryptodev_pmd.h   |  549 
 lib/librte_cryptodev/rte_cryptodev_version.map |   32 +
 lib/librte_eal/common/include/rte_log.h|1 +
 mk/rte.app.mk  |1 +
 14 files changed, 3021 insertions(+), 2 deletions(-)
 create mode 100644 lib/librte_cryptodev/Makefile
 create mode 100644 lib/librte_cryptodev/rte_crypto.h
 create mode 100644 lib/librte_cryptodev/rte_cryptodev.c
 create mode 100644 lib/librte_cryptodev/rte_cryptodev.h
 create mode 100644 lib/librte_cryptodev/rte_cryptodev_pmd.h
 create mode 100644 lib/librte_cryptodev/rte_cryptodev_version.map

diff --git a/MAINTAINERS b/MAINTAINERS
index d6feada..9138bbb 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -196,6 +196,10 @@ M: Thomas Monjalon 
 F: lib/librte_ether/
 F: scripts/test-null.sh

+Crypto API
+M: Declan Doherty 
+F: lib/librte_cryptodev
+F: docs/guides/cryptodevs

 Drivers
 ---
diff --git a/config/common_bsdapp b/config/common_bsdapp
index 7df0763..3bfb3f6 100644
--- a/config/common_bsdapp
+++ b/config/common_bsdapp
@@ -1,6 +1,6 @@
 #   BSD LICENSE
 #
-#   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+#   Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
 #   All rights reserved.
 #
 #   Redistribution and use in source and binary forms, with or without
@@ -150,6 +150,14 @@ CONFIG_RTE_ETHDEV_QUEUE_STAT_CNTRS=16
 CONFIG_RTE_ETHDEV_RXTX_CALLBACKS=y

 #
+# Compile generic Crypto device library
+#
+CONFIG_RTE_LIBRTE_CRYPTODEV=y
+CONFIG_RTE_LIBRTE_CRYPTODEV_DEBUG=n
+CONFIG_RTE_CRYPTO_MAX_DEVS=64
+CONFIG_RTE_CRYPTODEV_NAME_LEN=64
+
+#
 # Support NIC bypass logic
 #
 CONFIG_RTE_NIC_BYPASS=n
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 52173d5..cd7a2d4 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -1,6 +1,6 @@
 #   BSD LICENSE
 #
-#   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+#   Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
 #   All rights reserved.
 #
 #   Redistribution and use in source and binary forms, with or without
@@ -148,6 +148,14 @@ CONFIG_RTE_ETHDEV_QUEUE_STAT_CNTRS=16
 CONFIG_RTE_ETHDEV_RXTX_CALLBACKS=y

 #
+# Compile generic Crypto device library
+#
+CONFIG_RTE_LIBRTE_CRYPTODEV=y
+CONFIG_RTE_LIBRTE_CRYPTODEV_DEBUG=n
+CONFIG_RTE_CRYPTO_MAX_DEVS=64
+CONFIG_RTE_CRYPTODEV_NAME_LEN=64
+
+#
 # Support NIC bypass logic
 #
 CONFIG_RTE_NIC_BYPASS=n
diff --git a/doc/api/doxy-api-index.md b/doc/api/doxy-api-index.md
index 72ac3c4..bdb6130 100644
--- a/doc/api/doxy-api-index.md
+++ b/doc/api/doxy-api-index.md
@@ -39,6 +39,7 @@ There are many libraries, so their headers may be grouped by 
topics:
   [dev](@ref rte_dev.h),
   [ethdev] (@ref rte_ethdev.h),
   [ethctrl](@ref rte_eth_ctrl.h),
+  [cryptodev]  (@ref rte_cryptodev.h),
   [devargs](@ref rte_devargs.h),
   [bond]   (@ref rte_eth_bond.h),
   [vhost]  (@ref rte_virtio_net.h),
diff --git a/doc/api/doxy-api.conf b/doc/api/doxy-api.conf
index cfb4627..7244b8f 100644
--- a/doc/api/doxy-api.conf
+++ b/doc/api/doxy-api.conf
@@ -37,6 +37,7 @@ INPUT   = doc/api/doxy-

[dpdk-dev] [PATCH v8 07/10] qat_crypto_pmd: Addition of a new QAT DPDK PMD.

[Declan Doherty]

This patch adds a PMD for the Intel Quick Assist Technology DH895xxC
hardware accelerator.

This patch depends on a QAT PF driver for device initialization. See
the file docs/guides/cryptodevs/qat.rst for configuration details

This patch supports a limited subset of QAT device functionality,
currently supporting chaining of cipher and hash operations for the
following algorithmsd:

Cipher algorithms:
  - RTE_CRYPTO_CIPHER_AES_CBC (with 128-bit, 192-bit and 256-bit keys supported)

Hash algorithms:
  - RTE_CRYPTO_AUTH_SHA1_HMAC
  - RTE_CRYPTO_AUTH_SHA256_HMAC
  - RTE_CRYPTO_AUTH_SHA512_HMAC
  - RTE_CRYPTO_AUTH_AES_XCBC_MAC

Some limitation on this patchset which shall be contributed in a
subsequent release:
 - Chained mbufs are not supported.
 - Hash only is not supported.
 - Cipher only is not supported.
 - Only in-place is currently supported (destination address is
   the same as source address).
 - Only supports session-oriented API implementation (session-less
   APIs are not supported).

Signed-off-by: Declan Doherty 
Signed-off-by: John Griffin 
Signed-off-by: Des O Dea 
Signed-off-by: Fiona Trahe 

Acked-by: Sergio Gonzalez Monroy 

---
 config/common_bsdapp   |  14 +
 config/common_linuxapp |  14 +
 doc/guides/cryptodevs/index.rst|  38 ++
 doc/guides/cryptodevs/qat.rst  | 219 
 doc/guides/index.rst   |   1 +
 drivers/Makefile   |   1 +
 drivers/crypto/Makefile|  37 ++
 drivers/crypto/qat/Makefile|  63 +++
 .../qat/qat_adf/adf_transport_access_macros.h  | 174 ++
 drivers/crypto/qat/qat_adf/icp_qat_fw.h| 316 +++
 drivers/crypto/qat/qat_adf/icp_qat_fw_la.h | 404 ++
 drivers/crypto/qat/qat_adf/icp_qat_hw.h| 306 +++
 drivers/crypto/qat/qat_adf/qat_algs.h  | 125 +
 drivers/crypto/qat/qat_adf/qat_algs_build_desc.c   | 601 +
 drivers/crypto/qat/qat_crypto.c| 561 +++
 drivers/crypto/qat/qat_crypto.h| 124 +
 drivers/crypto/qat/qat_logs.h  |  78 +++
 drivers/crypto/qat/qat_qp.c| 429 +++
 drivers/crypto/qat/rte_pmd_qat_version.map |   3 +
 drivers/crypto/qat/rte_qat_cryptodev.c | 137 +
 mk/rte.app.mk  |   3 +
 21 files changed, 3648 insertions(+)
 create mode 100644 doc/guides/cryptodevs/index.rst
 create mode 100644 doc/guides/cryptodevs/qat.rst
 create mode 100644 drivers/crypto/Makefile
 create mode 100644 drivers/crypto/qat/Makefile
 create mode 100644 drivers/crypto/qat/qat_adf/adf_transport_access_macros.h
 create mode 100644 drivers/crypto/qat/qat_adf/icp_qat_fw.h
 create mode 100644 drivers/crypto/qat/qat_adf/icp_qat_fw_la.h
 create mode 100644 drivers/crypto/qat/qat_adf/icp_qat_hw.h
 create mode 100644 drivers/crypto/qat/qat_adf/qat_algs.h
 create mode 100644 drivers/crypto/qat/qat_adf/qat_algs_build_desc.c
 create mode 100644 drivers/crypto/qat/qat_crypto.c
 create mode 100644 drivers/crypto/qat/qat_crypto.h
 create mode 100644 drivers/crypto/qat/qat_logs.h
 create mode 100644 drivers/crypto/qat/qat_qp.c
 create mode 100644 drivers/crypto/qat/rte_pmd_qat_version.map
 create mode 100644 drivers/crypto/qat/rte_qat_cryptodev.c

diff --git a/config/common_bsdapp b/config/common_bsdapp
index e536fdf..3302d3f 100644
--- a/config/common_bsdapp
+++ b/config/common_bsdapp
@@ -158,6 +158,20 @@ CONFIG_RTE_CRYPTO_MAX_DEVS=64
 CONFIG_RTE_CRYPTODEV_NAME_LEN=64

 #
+# Compile PMD for QuickAssist based devices
+#
+CONFIG_RTE_LIBRTE_PMD_QAT=n
+CONFIG_RTE_LIBRTE_QAT_DEBUG_INIT=n
+CONFIG_RTE_LIBRTE_QAT_DEBUG_TX=n
+CONFIG_RTE_LIBRTE_QAT_DEBUG_RX=n
+CONFIG_RTE_LIBRTE_QAT_DEBUG_DRIVER=n
+#
+# Number of sessions to create in the session memory pool
+# on a single QuickAssist device.
+#
+CONFIG_RTE_MAX_QAT_SESSIONS=200
+
+#
 # Support NIC bypass logic
 #
 CONFIG_RTE_NIC_BYPASS=n
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 1947ce3..458b014 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -156,6 +156,20 @@ CONFIG_RTE_CRYPTO_MAX_DEVS=64
 CONFIG_RTE_CRYPTODEV_NAME_LEN=64

 #
+# Compile PMD for QuickAssist based devices
+#
+CONFIG_RTE_LIBRTE_PMD_QAT=n
+CONFIG_RTE_LIBRTE_PMD_QAT_DEBUG_INIT=n
+CONFIG_RTE_LIBRTE_PMD_QAT_DEBUG_TX=n
+CONFIG_RTE_LIBRTE_PMD_QAT_DEBUG_RX=n
+CONFIG_RTE_LIBRTE_PMD_QAT_DEBUG_DRIVER=n
+#
+# Number of sessions to create in the session memory pool
+# on a single QuickAssist device.
+#
+CONFIG_RTE_QAT_PMD_MAX_NB_SESSIONS=2048
+
+#
 # Support NIC bypass logic
 #
 CONFIG_RTE_NIC_BYPASS=n
diff --git a/doc/guides/cryptodevs/index.rst b/doc/guides/cryptodevs/index.rst
new file mode 100644
index 000..8ac928c
--- /dev/null
+++ b/doc/guides/cryptodevs/index.rst
@@ -0,0 +1,38 @@
+..  BS

[dpdk-dev] [PATCH v8 08/10] aesni_mb_pmd: Initial implementation of multi buffer based crypto device

[Declan Doherty]

This patch provides the initial implementation of the AES-NI multi-buffer
based crypto poll mode driver using DPDK's new cryptodev framework.

This PMD is dependent on Intel's multibuffer library, see the whitepaper
"Fast Multi-buffer IPsec Implementations on Intel? Architecture
Processors", see ref 1 for details on the library's design and ref 2 to
download the library itself. This initial implementation is limited to
supporting the chained operations of "hash then cipher" or "cipher then
hash" for the following cipher and hash algorithms:

Cipher algorithms:
  - RTE_CRYPTO_CIPHER_AES_CBC (with 128-bit, 192-bit and 256-bit keys supported)

Authentication algorithms:
  - RTE_CRYPTO_AUTH_SHA1_HMAC
  - RTE_CRYPTO_AUTH_SHA256_HMAC
  - RTE_CRYPTO_AUTH_SHA512_HMAC
  - RTE_CRYPTO_AUTH_AES_XCBC_MAC

Important Note:
Due to the fact that the multi-buffer library is designed for
accelerating IPsec crypto operation, the digest's generated for the HMAC
functions are truncated to lengths specified by IPsec RFC's, ie RFC2404
for using HMAC-SHA-1 with IPsec specifies that the digest is truncate
from 20 to 12 bytes.

Build instructions:
To build DPKD with the AESNI_MB_PMD the user is required to download
(ref 2) and compile the multi-buffer library on there system before
building DPDK. The environmental variable AESNI_MULTI_BUFFER_LIB_PATH
must be exported with the path where you extracted and built the multi
buffer library and finally set CONFIG_RTE_LIBRTE_PMD_AESNI_MB=y in
config/common_linuxapp.

Current status: It's doesn't support crypto operation
across chained mbufs, or cipher only or hash only operations.

ref 1:
https://www-ssl.intel.com/content/www/us/en/intelligent-systems/intel-technology/fast-multi-buffer-ipsec-implementations-ia-processors-p

ref 2: https://downloadcenter.intel.com/download/22972

Signed-off-by: Declan Doherty 
Acked-by: Sergio Gonzalez Monroy 

---
 MAINTAINERS|   3 +
 config/common_bsdapp   |   7 +
 config/common_linuxapp |   7 +
 doc/guides/cryptodevs/aesni_mb.rst |  85 +++
 doc/guides/cryptodevs/index.rst|   1 +
 drivers/crypto/Makefile|   1 +
 drivers/crypto/aesni_mb/Makefile   |  63 ++
 drivers/crypto/aesni_mb/aesni_mb_ops.h | 210 +++
 drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c | 669 +
 drivers/crypto/aesni_mb/rte_aesni_mb_pmd_ops.c | 298 +
 drivers/crypto/aesni_mb/rte_aesni_mb_pmd_private.h | 229 +++
 drivers/crypto/aesni_mb/rte_pmd_aesni_version.map  |   3 +
 mk/rte.app.mk  |   4 +
 13 files changed, 1580 insertions(+)
 create mode 100644 doc/guides/cryptodevs/aesni_mb.rst
 create mode 100644 drivers/crypto/aesni_mb/Makefile
 create mode 100644 drivers/crypto/aesni_mb/aesni_mb_ops.h
 create mode 100644 drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c
 create mode 100644 drivers/crypto/aesni_mb/rte_aesni_mb_pmd_ops.c
 create mode 100644 drivers/crypto/aesni_mb/rte_aesni_mb_pmd_private.h
 create mode 100644 drivers/crypto/aesni_mb/rte_pmd_aesni_version.map

diff --git a/MAINTAINERS b/MAINTAINERS
index dd8be0f..a51a660 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -305,6 +305,9 @@ Null PMD
 M: Tetsuya Mukawa 
 F: drivers/net/null/

+Crypto AES-NI Multi-Buffer PMD
+M: Declan Doherty 
+F: driver/crypto/aesni_mb

 Packet processing
 -
diff --git a/config/common_bsdapp b/config/common_bsdapp
index 3302d3f..6f0becb 100644
--- a/config/common_bsdapp
+++ b/config/common_bsdapp
@@ -171,6 +171,13 @@ CONFIG_RTE_LIBRTE_QAT_DEBUG_DRIVER=n
 #
 CONFIG_RTE_MAX_QAT_SESSIONS=200

+
+#
+# Compile PMD for AESNI backed device
+#
+CONFIG_RTE_LIBRTE_PMD_AESNI_MB=n
+CONFIG_RTE_LIBRTE_AESNI_MB_DEBUG=n
+
 #
 # Support NIC bypass logic
 #
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 458b014..ca6adc7 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -169,6 +169,13 @@ CONFIG_RTE_LIBRTE_PMD_QAT_DEBUG_DRIVER=n
 #
 CONFIG_RTE_QAT_PMD_MAX_NB_SESSIONS=2048

+# Compile PMD for AESNI backed device
+#
+CONFIG_RTE_LIBRTE_PMD_AESNI_MB=n
+CONFIG_RTE_LIBRTE_PMD_AESNI_MB_DEBUG=n
+CONFIG_RTE_AESNI_MB_PMD_MAX_NB_QUEUE_PAIRS=8
+CONFIG_RTE_AESNI_MB_PMD_MAX_NB_SESSIONS=2048
+
 #
 # Support NIC bypass logic
 #
diff --git a/doc/guides/cryptodevs/aesni_mb.rst 
b/doc/guides/cryptodevs/aesni_mb.rst
new file mode 100644
index 000..2ff5c41
--- /dev/null
+++ b/doc/guides/cryptodevs/aesni_mb.rst
@@ -0,0 +1,85 @@
+..  BSD LICENSE
+Copyright(c) 2015 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

[dpdk-dev] [PATCH v8 09/10] app/test: add cryptodev unit and performance tests

[Declan Doherty]

unit tests are run by using cryptodev_qat_autotest or
cryptodev_aesni_autotest from the test apps interactive console.

performance tests are run by using the cryptodev_qat_perftest or
cryptodev_aesni_mb_perftest command from the test apps interactive
console.

If you which to run the tests on a QAT device there must be one
bound to igb_uio kernel driver.

Signed-off-by: Declan Doherty 
Signed-off-by: John Griffin 
Signed-off-by: Des O Dea 
Signed-off-by: Fiona Trahe 

Acked-by: Sergio Gonzalez Monroy 

---
 MAINTAINERS  |2 +
 app/test/Makefile|4 +
 app/test/test.c  |   92 +-
 app/test/test.h  |   34 +-
 app/test/test_cryptodev.c| 1986 
 app/test/test_cryptodev.h|   68 ++
 app/test/test_cryptodev_perf.c   | 2062 ++
 app/test/test_link_bonding.c |6 +-
 app/test/test_link_bonding_mode4.c   |7 +-
 app/test/test_link_bonding_rssconf.c |7 +-
 10 files changed, 4219 insertions(+), 49 deletions(-)
 create mode 100644 app/test/test_cryptodev.c
 create mode 100644 app/test/test_cryptodev.h
 create mode 100644 app/test/test_cryptodev_perf.c

diff --git a/MAINTAINERS b/MAINTAINERS
index a51a660..74aa169 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -204,6 +204,8 @@ Crypto API
 M: Declan Doherty 
 F: lib/librte_cryptodev
 F: docs/guides/cryptodevs
+F: app/test/test_cryptodev.c
+F: app/test/test_cryptodev_perf.c

 Drivers
 ---
diff --git a/app/test/Makefile b/app/test/Makefile
index de63235..ec33e1a 100644
--- a/app/test/Makefile
+++ b/app/test/Makefile
@@ -149,6 +149,10 @@ endif

 SRCS-$(CONFIG_RTE_LIBRTE_PMD_RING) += test_pmd_ring.c
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_RING) += test_pmd_ring_perf.c
+
+SRCS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += test_cryptodev_perf.c
+SRCS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += test_cryptodev.c
+
 SRCS-$(CONFIG_RTE_LIBRTE_KVARGS) += test_kvargs.c

 CFLAGS += -O3
diff --git a/app/test/test.c b/app/test/test.c
index b94199a..f35b304 100644
--- a/app/test/test.c
+++ b/app/test/test.c
@@ -159,51 +159,81 @@ main(int argc, char **argv)
 int
 unit_test_suite_runner(struct unit_test_suite *suite)
 {
-   int retval, i = 0;
+   int test_success;
+   unsigned total = 0, executed = 0, skipped = 0, succeeded = 0, failed = 
0;

if (suite->suite_name)
-   printf("Test Suite : %s\n", suite->suite_name);
+   printf(" + 
--- +\n");
+   printf(" + Test Suite : %s\n", suite->suite_name);

if (suite->setup)
if (suite->setup() != 0)
-   return -1;
-
-   while (suite->unit_test_cases[i].testcase) {
-   /* Run test case setup */
-   if (suite->unit_test_cases[i].setup) {
-   retval = suite->unit_test_cases[i].setup();
-   if (retval != 0)
-   return retval;
-   }
+   goto suite_summary;

-   /* Run test case */
-   if (suite->unit_test_cases[i].testcase() == 0) {
-   printf("TestCase %2d: %s\n", i,
-   suite->unit_test_cases[i].success_msg ?
-   suite->unit_test_cases[i].success_msg :
-   "passed");
-   }
-   else {
-   printf("TestCase %2d: %s\n", i, 
suite->unit_test_cases[i].fail_msg ?
-   suite->unit_test_cases[i].fail_msg :
-   "failed");
-   return -1;
+   printf(" + --- 
+\n");
+
+   while (suite->unit_test_cases[total].testcase) {
+   if (!suite->unit_test_cases[total].enabled) {
+   skipped++;
+   total++;
+   continue;
+   } else {
+   executed++;
}

-   /* Run test case teardown */
-   if (suite->unit_test_cases[i].teardown) {
-   retval = suite->unit_test_cases[i].teardown();
-   if (retval != 0)
-   return retval;
+   /* run test case setup */
+   if (suite->unit_test_cases[total].setup)
+   test_success = suite->unit_test_cases[total].setup();
+   else
+   test_success = TEST_SUCCESS;
+
+   if (test_success == TEST_SUCCESS) {
+   /* run the test case */
+   test_success = suite->unit_test_cases[total].testcase();
+   if (test_success == TEST_SUCCESS)
+   succeeded++;
+   else
+  

[dpdk-dev] [PATCH v8 10/10] l2fwd-crypto: crypto

[Declan Doherty]

This patch creates a new sample applicaiton based off the l2fwd
application which performs specified crypto operations on IP packet
payloads which are forwarding.

Signed-off-by: Declan Doherty 
Acked-by: Sergio Gonzalez Monroy 

---
 MAINTAINERS|1 +
 examples/Makefile  |1 +
 examples/l2fwd-crypto/Makefile |   50 ++
 examples/l2fwd-crypto/main.c   | 1489 
 4 files changed, 1541 insertions(+)
 create mode 100644 examples/l2fwd-crypto/Makefile
 create mode 100644 examples/l2fwd-crypto/main.c

diff --git a/MAINTAINERS b/MAINTAINERS
index 74aa169..0685a56 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -206,6 +206,7 @@ F: lib/librte_cryptodev
 F: docs/guides/cryptodevs
 F: app/test/test_cryptodev.c
 F: app/test/test_cryptodev_perf.c
+F: examples/l2fwd-crypto

 Drivers
 ---
diff --git a/examples/Makefile b/examples/Makefile
index 830e31a..2da9b49 100644
--- a/examples/Makefile
+++ b/examples/Makefile
@@ -75,5 +75,6 @@ DIRS-$(CONFIG_RTE_LIBRTE_XEN_DOM0) += vhost_xen
 DIRS-y += vmdq
 DIRS-y += vmdq_dcb
 DIRS-$(CONFIG_RTE_LIBRTE_POWER) += vm_power_manager
+DIRS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += l2fwd-crypto

 include $(RTE_SDK)/mk/rte.extsubdir.mk
diff --git a/examples/l2fwd-crypto/Makefile b/examples/l2fwd-crypto/Makefile
new file mode 100644
index 000..e8224ca
--- /dev/null
+++ b/examples/l2fwd-crypto/Makefile
@@ -0,0 +1,50 @@
+#   BSD LICENSE
+#
+#   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+#   All rights reserved.
+#
+#   Redistribution and use in source and binary forms, with or without
+#   modification, are permitted provided that the following conditions
+#   are met:
+#
+# * Redistributions of source code must retain the above copyright
+#   notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above copyright
+#   notice, this list of conditions and the following disclaimer in
+#   the documentation and/or other materials provided with the
+#   distribution.
+# * Neither the name of Intel Corporation nor the names of its
+#   contributors may be used to endorse or promote products derived
+#   from this software without specific prior written permission.
+#
+#   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+#   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+#   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+#   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+#   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+#   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+#   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+#   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+#   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+#   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ifeq ($(RTE_SDK),)
+$(error "Please define RTE_SDK environment variable")
+endif
+
+# Default target, can be overridden by command line or environment
+RTE_TARGET ?= x86_64-native-linuxapp-gcc
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+# binary name
+APP = l2fwd-crypto
+
+# all source are stored in SRCS-y
+SRCS-y := main.c
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+include $(RTE_SDK)/mk/rte.extapp.mk
diff --git a/examples/l2fwd-crypto/main.c b/examples/l2fwd-crypto/main.c
new file mode 100644
index 000..0b4414b
--- /dev/null
+++ b/examples/l2fwd-crypto/main.c
@@ -0,0 +1,1489 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2010-2014 Intel Corporation. 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 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, EXEMP

[dpdk-dev] [PATCH] sched: remove unused debug functions

[Thomas Monjalon]

2015-11-25 01:52, Thomas Monjalon:
> clang warns about unused functions when debug is enabled.
> 
> Signed-off-by: Thomas Monjalon 

Applied

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Thomas Monjalon]

2015-11-25 11:13, Ferruh Yigit:
> Fixes: 67b6d3039e9e ("eal: set name to threads")
> 
> pthread_setname_np() function added in glibc 2.12, using this function
> in older glibc versions cause compile error:
> error: implicit declaration of function "pthread_setname_np"
> 
> This patch adds "rte_thread_setname" macro and set it according
> glibc >= 2.12 check, thread naming disabled for older glibc versions,
> glibc versions that support "pthread_setname_np" will keep using this
> function.
> 
> Signed-off-by: Ferruh Yigit 

Applied as eal/linux fix, thanks

[dpdk-dev] |ERROR| pw 9061-9064 mk: add xgene1 machine target based on armv8-a

[Jerin Jacob]

On Mon, Nov 23, 2015 at 11:34:49AM -0800, sys_stv at intel.com wrote:
> Test-Label: Intel Niantic on Fedora
> Test-Status: ERROR
> Patchwork: http://www.dpdk.org/dev/patchwork/patch/9064/
> 
> DPDK git baseline: 46454f16d548afc202f157751365088910dcf0e7
> Patchwork ID: 9061-9064
> http://www.dpdk.org/dev/patchwork/patch/9064/
> Submitter: Jerin Jacob 
> Date: Tue, 24 Nov 2015 00:15:38 +0530
> 
> Source Compilation:
> OS: fedora
> Nic: niantic
> i686-native-linuxapp-gcc: compile pass
> x86_64-native-linuxapp-gcc: compile pass
> 
> DTS validation: 
> OS: fedora
> Nic: Niantic
> TestType: auto
> GCC: 4
> x86_64-native-linuxapp-gcc:  total 79, passed 78, failed 1.
> Failed Case List:
> Target: x86_64-native-linuxapp-gcc
> OS: fedora
> Failed DTS case: 
> link_bonding: 
> http://dpdk.org/browse/tools/dts/tree/test_plans/unit_tests_eal_test_plan.rst
> 
> DTS Validation Error:
> 
> 
> TEST SUITE : TestUnitTestsEal
> [SUITE_DUT_CMD] make -j -C ./app/test/ 
> 
> ---
> Begin: Test Casetest_link_bonding
> --
> FAILED  'Test failed'

I received this auto email regarding 'Test failed' and
it is referring to this patch.  I don't think I've changed anything
that could cause this.  Kindly advise.



> --
> [SUITE_DUT_CMD]  ./app/test/test -n 1 -c 
> [SUITE_DUT_CMD]  link_bonding_autotest
> [SUITE_DUT_CMD]  quit
> End test_link_bonding
> ---
> 
> 
> 
> DPDK STV team 

[dpdk-dev] no hugepage with UIO poll-mode driver

[Sergio Gonzalez Monroy]

On 25/11/2015 13:22, Thomas Monjalon wrote:
> 2015-11-25 12:02, Bruce Richardson:
>> On Wed, Nov 25, 2015 at 12:03:05PM +0100, Thomas Monjalon wrote:
>>> 2015-11-25 11:00, Bruce Richardson:
 On Wed, Nov 25, 2015 at 11:23:57AM +0100, Thomas Monjalon wrote:
> 2015-11-25 10:08, Bruce Richardson:
>> On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
>>> Hi Jianfeng,
>>>
>>> Thanks for the email. rte mempool was successfully created without any
>>> error. Now the next problem is that rte_eth_rx_burst() is always 
>>> returning 0
>>> as if there was no packet to receive... Do you have any suggestion on 
>>> what
>>> might be causing this issue? In the meantime, I will be digging through
>>> ixgbe driver code to see what's going on.
>>>
>>> Thank you,
>>> Younghwan
>>>
>> The problem is that with --no-huge we don't have the physical address of 
>> the memory
>> to write to the network card. That's what it's marked as for testing 
>> only.
> Even with rte_mem_virt2phy() + rte_mem_lock_page() ?
>
 With no-huge, we just set up a single memory segment at startup and set its
 "physaddr" to be the virtual address.

  /* hugetlbfs can be disabled */
  if (internal_config.no_hugetlbfs) {
  addr = mmap(NULL, internal_config.memory, PROT_READ | 
 PROT_WRITE,
  MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
  if (addr == MAP_FAILED) {
  RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", 
 __func__,
  strerror(errno));
  return -1;
  }
  mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr;
>>> rte_mem_virt2phy() does not use memseg.phys_addr but /proc/self/pagemap:
>>>
>>>  /*
>>>   * the pfn (page frame number) are bits 0-54 (see
>>>   * pagemap.txt in linux Documentation)
>>>   */
>>>  physaddr = ((page & 0x7fULL) * page_size)
>>>  + ((unsigned long)virtaddr % page_size);
>>>
>> Yes, you are right. I was not aware that that function was used as part of 
>> the
>> mempool init, but now I see that "rte_mempool_virt2phy()" does indeed call 
>> that
>> function if hugepages are disabled, so my bad.
> Do you think we could move --no-huge in the main section (not only for 
> testing)?
Hi,

I think the main issue is going to be the HW descriptors queues.
AFAIK drivers now call rte_eth_dma_zone_reserve, which is basically a 
wrapper around
rte_memzone_reserve, to get a chunk of phys memory, and in the case of 
--no-huge is
not going to be really phys contiguous.

Ideally we would move and expand the functionality of dma_zone reserve 
API to the EAL,
so we could detect what page size we have and set the boundary for such 
page size.
dma_zone_reserve does something similar to work on Xen target by 
reserving memzones
on 2MB boundary.

Sergio

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Roger B. Melton]

> +/**
> + * Set thread names.
> + *
> + * Macro to wrap `pthread_setname_np()` with a glibc version check.
> + * Only glibc >= 2.12 supports this feature.
> + *
> + * This macro only used for Linux, BSD does direct libc call.
> + * BSD libc version of function is `pthread_set_name_np()`.
> + */
> +#if defined(__DOXYGEN__)
> +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> +#endif
> +
> +#if defined(__GLIBC__) && defined(__GLIBC_PREREQ)
> +#if __GLIBC_PREREQ(2, 12)
> +#define rte_thread_setname(...) pthread_setname_np(__VA_ARGS__)
> +#else
> +#define rte_thread_setname(...) 0
> +#endif
> +#endif

Have you thought about a way to set thread name when glibc < 2.12.  I 
also ran into the problem recently and played around with prctl() 
(Linux) to set thread (process) name.  e.g.

ret = prctl(PR_SET_NAME,,0,0,0);


There are 2 issues I think:

1) The semantics are different than prthread_setname_np().  With
pthread_setname_np() a name can be assigned to any thread, with
prctl() the name is assigned to the active thread.  That would mean
that rather than rte_eal_init(), rte_eal_intr_init() could not
assign thread names.  Rather the threads would have to name themselves.

2) I think BSD lacks prctl(), but some (not all?) BSD
implementations have setproctitle() to do the same thing.


It might be too late for 2.2, but something to think about for the future.

Regards,
Roger

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Thomas Monjalon]

2015-11-25 08:51, Roger B. Melton:
> Have you thought about a way to set thread name when glibc < 2.12.  I 
> also ran into the problem recently and played around with prctl() 
> (Linux) to set thread (process) name.  e.g.
> 
> ret = prctl(PR_SET_NAME,,0,0,0);
> 
> 
> There are 2 issues I think:
> 
> 1) The semantics are different than prthread_setname_np().  With
> pthread_setname_np() a name can be assigned to any thread, with
> prctl() the name is assigned to the active thread.  That would mean
> that rather than rte_eal_init(), rte_eal_intr_init() could not
> assign thread names.  Rather the threads would have to name themselves.
> 
> 2) I think BSD lacks prctl(), but some (not all?) BSD
> implementations have setproctitle() to do the same thing.
> 
> 
> It might be too late for 2.2, but something to think about for the future.

I don't think this feature is important enough to deal with old environments
and to risk some complicated bugs.
Do you think it deserves more tricks?

[dpdk-dev] no hugepage with UIO poll-mode driver

[Ananyev, Konstantin]

> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Sergio Gonzalez Monroy
> Sent: Wednesday, November 25, 2015 1:44 PM
> To: Thomas Monjalon
> Cc: dev at dpdk.org
> Subject: Re: [dpdk-dev] no hugepage with UIO poll-mode driver
> 
> On 25/11/2015 13:22, Thomas Monjalon wrote:
> > 2015-11-25 12:02, Bruce Richardson:
> >> On Wed, Nov 25, 2015 at 12:03:05PM +0100, Thomas Monjalon wrote:
> >>> 2015-11-25 11:00, Bruce Richardson:
>  On Wed, Nov 25, 2015 at 11:23:57AM +0100, Thomas Monjalon wrote:
> > 2015-11-25 10:08, Bruce Richardson:
> >> On Wed, Nov 25, 2015 at 03:39:17PM +0900, Younghwan Go wrote:
> >>> Hi Jianfeng,
> >>>
> >>> Thanks for the email. rte mempool was successfully created without any
> >>> error. Now the next problem is that rte_eth_rx_burst() is always 
> >>> returning 0
> >>> as if there was no packet to receive... Do you have any suggestion on 
> >>> what
> >>> might be causing this issue? In the meantime, I will be digging 
> >>> through
> >>> ixgbe driver code to see what's going on.
> >>>
> >>> Thank you,
> >>> Younghwan
> >>>
> >> The problem is that with --no-huge we don't have the physical address 
> >> of the memory
> >> to write to the network card. That's what it's marked as for testing 
> >> only.
> > Even with rte_mem_virt2phy() + rte_mem_lock_page() ?
> >
>  With no-huge, we just set up a single memory segment at startup and set 
>  its
>  "physaddr" to be the virtual address.
> 
>   /* hugetlbfs can be disabled */
>   if (internal_config.no_hugetlbfs) {
>   addr = mmap(NULL, internal_config.memory, PROT_READ | 
>  PROT_WRITE,
>   MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
>   if (addr == MAP_FAILED) {
>   RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", 
>  __func__,
>   strerror(errno));
>   return -1;
>   }
>   mcfg->memseg[0].phys_addr = 
>  (phys_addr_t)(uintptr_t)addr;
> >>> rte_mem_virt2phy() does not use memseg.phys_addr but /proc/self/pagemap:
> >>>
> >>>  /*
> >>>   * the pfn (page frame number) are bits 0-54 (see
> >>>   * pagemap.txt in linux Documentation)
> >>>   */
> >>>  physaddr = ((page & 0x7fULL) * page_size)
> >>>  + ((unsigned long)virtaddr % page_size);
> >>>
> >> Yes, you are right. I was not aware that that function was used as part of 
> >> the
> >> mempool init, but now I see that "rte_mempool_virt2phy()" does indeed call 
> >> that
> >> function if hugepages are disabled, so my bad.
> > Do you think we could move --no-huge in the main section (not only for 
> > testing)?
> Hi,
> 
> I think the main issue is going to be the HW descriptors queues.
> AFAIK drivers now call rte_eth_dma_zone_reserve, which is basically a
> wrapper around
> rte_memzone_reserve, to get a chunk of phys memory, and in the case of
> --no-huge is
> not going to be really phys contiguous.
> 
> Ideally we would move and expand the functionality of dma_zone reserve
> API to the EAL,
> so we could detect what page size we have and set the boundary for such
> page size.
> dma_zone_reserve does something similar to work on Xen target by
> reserving memzones
> on 2MB boundary.

With xen we have a special kernel driver that allocates physically continuous 
chunks of memory for us.
So we can guarantee that each such chunk would be at least 2MB long.
That's enough to allocate HW rings (max HW ring size for let say ixgbe is 
~64KB).
Here there is absolutely no guarantee that memory allocated by kernel will be 
memory continuous.
Of course you can search though all pages that you allocated and most likely 
you'll find a continuous
chunk big enough for that.
Another problem - mbufs. 
You need to be sure that each mbuf doesn't cross page boundary
(in case next page is not adjacent to current one).
So you'll probably need to use rte_mempool_xmem_create() to allocate mbufs from 
no hugepages.
BTW, as I remember with vfio in place you should be able to do IO with 
no-hugepages options, no?
As it relies on vfio ability to setup IOMMU tables for you.
Konstantin

> 
> Sergio

[dpdk-dev] [PATCH v2] eal: fix compile error for old glibc caused by pthread_setname_np()

[Roger B. Melton]

On 11/25/15 9:03 AM, Thomas Monjalon wrote:
> 2015-11-25 08:51, Roger B. Melton:
>> Have you thought about a way to set thread name when glibc < 2.12.  I
>> also ran into the problem recently and played around with prctl()
>> (Linux) to set thread (process) name.  e.g.
>>
>>  ret = prctl(PR_SET_NAME,,0,0,0);
>>
>>
>> There are 2 issues I think:
>>
>>  1) The semantics are different than prthread_setname_np().  With
>>  pthread_setname_np() a name can be assigned to any thread, with
>>  prctl() the name is assigned to the active thread.  That would mean
>>  that rather than rte_eal_init(), rte_eal_intr_init() could not
>>  assign thread names.  Rather the threads would have to name themselves.
>>
>>  2) I think BSD lacks prctl(), but some (not all?) BSD
>>  implementations have setproctitle() to do the same thing.
>>
>>
>> It might be too late for 2.2, but something to think about for the future.
> I don't think this feature is important enough to deal with old environments
> and to risk some complicated bugs.
> Do you think it deserves more tricks?
> .
>
I agree with you Thomas.  While I am one of those living in an old 
environment, I believe that the complications of the tricks out weight 
the debug benefit.  However there may be other in the community who have 
a different view, so I thought I would at least suggest that there are 
alternatives.

Thanks,
-Roger

[dpdk-dev] [PATCH] doc: update docs for minimum kernel requirement.

[John McNamara]

Update the documentation to reflect that the minimum Linux kernel
requirement for DPDK 2.2 has increased from 2.6.33 to 2.6.34.

Compatibility with kernel 2.6.33 was dropped, after discussion on
the mailing list, in the following commit:

Commit: 2e6e9e215703 ("igb_uio: use existing PCI macros")

Signed-off-by: John McNamara 
---

Original discussions leading to this patch:

http://thread.gmane.org/gmane.comp.networking.dpdk.devel/23523

 doc/guides/faq/faq.rst   | 19 ---
 doc/guides/linux_gsg/sys_reqs.rst|  6 +-
 doc/guides/rel_notes/release_2_2.rst |  3 +++
 3 files changed, 4 insertions(+), 24 deletions(-)

diff --git a/doc/guides/faq/faq.rst b/doc/guides/faq/faq.rst
index 0686e17..d58673b 100644
--- a/doc/guides/faq/faq.rst
+++ b/doc/guides/faq/faq.rst
@@ -160,25 +160,6 @@ This is especially true when using a large number (>512) 
of 2 MB huge pages. Ple
 This can be done either by issuing a ulimit command or editing the limits.conf 
file. Please consult Linux* manpages for usage information.


-Does my kernel require patching to run the?DPDK?
-
-
-Any kernel greater than version 2.6.33 can be used without any patches 
applied. The following kernels may require patches to provide hugepage support:
-
-*   Kernel version 2.6.32 requires the following patches applied:
-
-*   `mm: hugetlb: add hugepage support to pagemap 
`_
-
-*   `mm: hugetlb: fix hugepage memory leak in walk_page_range() 
`_
-
-*   `hugetlb: add nodemask arg to huge page alloc, free and surplus adjust 
functions 
`_
-(not mandatory, but recommended on a NUMA system to support per-NUMA 
node hugepages allocation)
-
-*   Kernel version 2.6.31, requires the above patches plus the following:
-
-*   `UIO: Add name attributes for mappings and port regions 
`_
-
-
 VF driver for IXGBE devices cannot be initialized
 -

diff --git a/doc/guides/linux_gsg/sys_reqs.rst 
b/doc/guides/linux_gsg/sys_reqs.rst
index ebed418..cf05983 100644
--- a/doc/guides/linux_gsg/sys_reqs.rst
+++ b/doc/guides/linux_gsg/sys_reqs.rst
@@ -111,7 +111,7 @@ System Software

 **Required:**

-*   Kernel version >= 2.6.33
+*   Kernel version >= 2.6.34

 The kernel version in use can be checked using the command:

@@ -119,10 +119,6 @@ System Software

 uname -r

-For details of the patches needed to use the DPDK with earlier kernel versions,
-see the DPDK FAQ included in the *DPDK Release Notes*.
-Note also that Red hat* Linux* 6.2 and 6.3 uses a 2.6.32 kernel that already 
has all the necessary patches applied.
-
 *   glibc >= 2.7 (for features related to cpuset)

 The version can be checked using the ldd --version command. A sample 
output is shown below:
diff --git a/doc/guides/rel_notes/release_2_2.rst 
b/doc/guides/rel_notes/release_2_2.rst
index 8c77768..cd5c5e9 100644
--- a/doc/guides/rel_notes/release_2_2.rst
+++ b/doc/guides/rel_notes/release_2_2.rst
@@ -224,6 +224,9 @@ Examples
 Other
 ~

+* This release drops compatibility with Linux kernel 2.6.33. The minimum
+  kernel requirement is now 2.6.34.
+

 Known Issues
 
-- 
2.5.0

[dpdk-dev] [PATCH v7 4/9] nfp: adding stats

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  179 +
 1 file changed, 179 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index e40b613..49c71bf 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -90,6 +90,9 @@ static int nfp_net_tx_queue_setup(struct rte_eth_dev *dev, 
uint16_t queue_idx,
  uint16_t nb_desc, unsigned int socket_id,
  const struct rte_eth_txconf *tx_conf);
 static int nfp_net_start(struct rte_eth_dev *dev);
+static void nfp_net_stats_get(struct rte_eth_dev *dev,
+ struct rte_eth_stats *stats);
+static void nfp_net_stats_reset(struct rte_eth_dev *dev);
 static void nfp_net_stop(struct rte_eth_dev *dev);
 static uint16_t nfp_net_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
  uint16_t nb_pkts);
@@ -679,6 +682,177 @@ nfp_net_close(struct rte_eth_dev *dev)
 */
 }

+static void
+nfp_net_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
+{
+   int i;
+   struct nfp_net_hw *hw;
+   struct rte_eth_stats nfp_dev_stats;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   /* RTE_ETHDEV_QUEUE_STAT_CNTRS default value is 16 */
+
+   /* reading per RX ring stats */
+   for (i = 0; i < dev->data->nb_rx_queues; i++) {
+   if (i == RTE_ETHDEV_QUEUE_STAT_CNTRS)
+   break;
+
+   nfp_dev_stats.q_ipackets[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_RXR_STATS(i));
+
+   nfp_dev_stats.q_ipackets[i] -=
+   hw->eth_stats_base.q_ipackets[i];
+
+   nfp_dev_stats.q_ibytes[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_RXR_STATS(i) + 0x8);
+
+   nfp_dev_stats.q_ibytes[i] -=
+   hw->eth_stats_base.q_ibytes[i];
+   }
+
+   /* reading per TX ring stats */
+   for (i = 0; i < dev->data->nb_tx_queues; i++) {
+   if (i == RTE_ETHDEV_QUEUE_STAT_CNTRS)
+   break;
+
+   nfp_dev_stats.q_opackets[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_TXR_STATS(i));
+
+   nfp_dev_stats.q_opackets[i] -=
+   hw->eth_stats_base.q_opackets[i];
+
+   nfp_dev_stats.q_obytes[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_TXR_STATS(i) + 0x8);
+
+   nfp_dev_stats.q_obytes[i] -=
+   hw->eth_stats_base.q_obytes[i];
+   }
+
+   nfp_dev_stats.ipackets =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_FRAMES);
+
+   nfp_dev_stats.ipackets -= hw->eth_stats_base.ipackets;
+
+   nfp_dev_stats.ibytes =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_OCTETS);
+
+   nfp_dev_stats.ibytes -= hw->eth_stats_base.ibytes;
+
+   nfp_dev_stats.opackets =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_TX_FRAMES);
+
+   nfp_dev_stats.opackets -= hw->eth_stats_base.opackets;
+
+   nfp_dev_stats.obytes =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_TX_OCTETS);
+
+   nfp_dev_stats.obytes -= hw->eth_stats_base.obytes;
+
+   nfp_dev_stats.imcasts =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_MC_FRAMES);
+
+   nfp_dev_stats.imcasts -= hw->eth_stats_base.imcasts;
+
+   /* reading general device stats */
+   nfp_dev_stats.ierrors =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_ERRORS);
+
+   nfp_dev_stats.ierrors -= hw->eth_stats_base.ierrors;
+
+   nfp_dev_stats.oerrors =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_TX_ERRORS);
+
+   nfp_dev_stats.oerrors -= hw->eth_stats_base.oerrors;
+
+   /* Multicast frames received */
+   nfp_dev_stats.imcasts =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_MC_FRAMES);
+
+   nfp_dev_stats.imcasts -= hw->eth_stats_base.imcasts;
+
+   /* RX ring mbuf allocation failures */
+   nfp_dev_stats.rx_nombuf = dev->data->rx_mbuf_alloc_failed;
+
+   nfp_dev_stats.imissed =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_DISCARDS);
+
+   nfp_dev_stats.imissed -= hw->eth_stats_base.imissed;
+
+   if (stats)
+   memcpy(stats, &nfp_dev_stats, sizeof(*stats));
+}
+
+static void
+nfp_net_stats_reset(struct rte_eth_dev *dev)
+{
+   int i;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   /*
+* hw->eth_stats_base records the per counter starting point.
+* Lets update it now
+*/
+
+   /* reading per RX ring stats */
+   for (i = 0; i < dev->data->nb_rx_queues; i++) {
+   if (i == RTE_ETHDEV_QUEUE_STAT_CNTRS)
+   break;
+
+   hw->eth_stats_base.q_ipackets[i] =
+   nn_cfg_rea

[dpdk-dev] [PATCH v7 0/9] support for netronome nfp-6xxx card

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

This patchset adds a new PMD for Netronome nfp-6xxx card.
Just PCI Virtual Functions supported.
Using this PMD requires previous Netronome BSP installation.

v7:
 - Adding support for link status changes interrupts
 - removing unnecessary flags when compiling the PMD

v6:
 - Making each patch compilable

v5:
 - Splitting up patches per functionality.

v4:
 - Getting rid of nfp_uio. Just submitting PMD.

v3:
 - Making all patches independent for applying and building.
 - changing commits messages following standard.

v2:
 - Code style changes based on checkpatch.pl and DPDK style guide.
 - Documentation changes using the right rst format.
 - Moving the documentation files to a new patch file.
 - Adding info to MAINTAINERS and release files.

*** BLURB HERE ***

Alejandro.Lucero (9):
  nfp: basic initialization
  nfp: adding rx/tx functionality
  nfp: adding rss
  nfp: adding stats
  nfp: adding link functionality
  nfp: adding extra functionality
  nfp: link status change interrupt support
  nfp: adding nic guide
  nfp: Updating MAINTAINERS

 MAINTAINERS  |4 +
 config/common_linuxapp   |6 +
 doc/guides/nics/index.rst|1 +
 doc/guides/nics/nfp.rst  |  265 
 doc/guides/rel_notes/release_2_2.rst |4 +
 drivers/net/Makefile |1 +
 drivers/net/nfp/Makefile |   57 +
 drivers/net/nfp/nfp_net.c| 2499 ++
 drivers/net/nfp/nfp_net_ctrl.h   |  324 +
 drivers/net/nfp/nfp_net_logs.h   |   75 +
 drivers/net/nfp/nfp_net_pmd.h|  453 ++
 mk/rte.app.mk|1 +
 12 files changed, 3690 insertions(+)
 create mode 100644 doc/guides/nics/nfp.rst
 create mode 100644 drivers/net/nfp/Makefile
 create mode 100644 drivers/net/nfp/nfp_net.c
 create mode 100644 drivers/net/nfp/nfp_net_ctrl.h
 create mode 100644 drivers/net/nfp/nfp_net_logs.h
 create mode 100644 drivers/net/nfp/nfp_net_pmd.h

-- 
1.7.9.5

[dpdk-dev] [PATCH v7 3/9] nfp: adding rss

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  218 +
 1 file changed, 218 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index fe62b4e..e40b613 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -1501,12 +1501,230 @@ xmit_end:
return i;
 }

+/* Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device */
+static int
+nfp_net_reta_update(struct rte_eth_dev *dev,
+   struct rte_eth_rss_reta_entry64 *reta_conf,
+   uint16_t reta_size)
+{
+   uint32_t reta, mask;
+   int i, j;
+   int idx, shift;
+   uint32_t update;
+   struct nfp_net_hw *hw =
+   NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (!(hw->ctrl & NFP_NET_CFG_CTRL_RSS))
+   return -EINVAL;
+
+   if (reta_size != NFP_NET_CFG_RSS_ITBL_SZ) {
+   RTE_LOG(ERR, PMD, "The size of hash lookup table configured "
+   "(%d) doesn't match the number hardware can supported "
+   "(%d)\n", reta_size, NFP_NET_CFG_RSS_ITBL_SZ);
+   return -EINVAL;
+   }
+
+   /*
+* Update Redirection Table. There are 128 8bit-entries which can be
+* manage as 32 32bit-entries
+*/
+   for (i = 0; i < reta_size; i += 4) {
+   /* Handling 4 RSS entries per loop */
+   idx = i / RTE_RETA_GROUP_SIZE;
+   shift = i % RTE_RETA_GROUP_SIZE;
+   mask = (uint8_t)((reta_conf[idx].mask >> shift) & 0xF);
+
+   if (!mask)
+   continue;
+
+   reta = 0;
+   /* If all 4 entries were set, don't need read RETA register */
+   if (mask != 0xF)
+   reta = nn_cfg_readl(hw, NFP_NET_CFG_RSS_ITBL + i);
+
+   for (j = 0; j < 4; j++) {
+   if (!(mask & (0x1 << j)))
+   continue;
+   if (mask != 0xF)
+   /* Clearing the entry bits */
+   reta &= ~(0xFF << (8 * j));
+   reta |= reta_conf[idx].reta[shift + j] << (8 * j);
+   }
+   nn_cfg_writel(hw, NFP_NET_CFG_RSS_ITBL + shift, reta);
+   }
+
+   update = NFP_NET_CFG_UPDATE_RSS;
+
+   if (nfp_net_reconfig(hw, hw->ctrl, update) < 0)
+   return -EIO;
+
+   return 0;
+}
+
+ /* Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device. 
*/
+static int
+nfp_net_reta_query(struct rte_eth_dev *dev,
+  struct rte_eth_rss_reta_entry64 *reta_conf,
+  uint16_t reta_size)
+{
+   uint8_t i, j, mask;
+   int idx, shift;
+   uint32_t reta;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (!(hw->ctrl & NFP_NET_CFG_CTRL_RSS))
+   return -EINVAL;
+
+   if (reta_size != NFP_NET_CFG_RSS_ITBL_SZ) {
+   RTE_LOG(ERR, PMD, "The size of hash lookup table configured "
+   "(%d) doesn't match the number hardware can supported "
+   "(%d)\n", reta_size, NFP_NET_CFG_RSS_ITBL_SZ);
+   return -EINVAL;
+   }
+
+   /*
+* Reading Redirection Table. There are 128 8bit-entries which can be
+* manage as 32 32bit-entries
+*/
+   for (i = 0; i < reta_size; i += 4) {
+   /* Handling 4 RSS entries per loop */
+   idx = i / RTE_RETA_GROUP_SIZE;
+   shift = i % RTE_RETA_GROUP_SIZE;
+   mask = (uint8_t)((reta_conf[idx].mask >> shift) & 0xF);
+
+   if (!mask)
+   continue;
+
+   reta = nn_cfg_readl(hw, NFP_NET_CFG_RSS_ITBL + shift);
+   for (j = 0; j < 4; j++) {
+   if (!(mask & (0x1 << j)))
+   continue;
+   reta_conf->reta[shift + j] =
+   (uint8_t)((reta >> (8 * j)) & 0xF);
+   }
+   }
+   return 0;
+}
+
+static int
+nfp_net_rss_hash_update(struct rte_eth_dev *dev,
+   struct rte_eth_rss_conf *rss_conf)
+{
+   uint32_t update;
+   uint32_t cfg_rss_ctrl = 0;
+   uint8_t key;
+   uint64_t rss_hf;
+   int i;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   rss_hf = rss_conf->rss_hf;
+
+   /* Checking if RSS is enabled */
+   if (!(hw->ctrl & NFP_NET_CFG_CTRL_RSS)) {
+   if (rss_hf != 0) { /* Enable RSS? */
+   RTE_LOG(ERR, PMD, "RSS unsupported\n");
+   return -EINVAL;
+   }
+   return 0; /* Nothing to do */
+   }
+
+   if (rss_conf->rss_key

[dpdk-dev] [PATCH v7 5/9] nfp: adding link functionality

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |   96 +
 1 file changed, 96 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index 49c71bf..a7db4bc 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -74,6 +74,7 @@
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
 static int nfp_net_init(struct rte_eth_dev *eth_dev);
+static int nfp_net_link_update(struct rte_eth_dev *dev, int wait_to_complete);
 static int nfp_net_rx_fill_freelist(struct nfp_net_rxq *rxq);
 static uint32_t nfp_net_rx_queue_count(struct rte_eth_dev *dev,
   uint16_t queue_idx);
@@ -226,6 +227,57 @@ ring_dma_zone_reserve(struct rte_eth_dev *dev, const char 
*ring_name,
   NFP_MEMZONE_ALIGN);
 }

+/*
+ * Atomically reads 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
+nfp_net_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
+nfp_net_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
 nfp_net_rx_queue_release_mbufs(struct nfp_net_rxq *rxq)
 {
@@ -682,6 +734,49 @@ nfp_net_close(struct rte_eth_dev *dev)
 */
 }

+/*
+ * return 0 means link status changed, -1 means not changed
+ *
+ * Wait to complete is needed as it can take up to 9 seconds to get the Link
+ * status.
+ */
+static int
+nfp_net_link_update(struct rte_eth_dev *dev, __rte_unused int wait_to_complete)
+{
+   struct nfp_net_hw *hw;
+   struct rte_eth_link link, old;
+   uint32_t nn_link_status;
+
+   PMD_DRV_LOG(DEBUG, "Link update\n");
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   memset(&old, 0, sizeof(old));
+   nfp_net_dev_atomic_read_link_status(dev, &old);
+
+   nn_link_status = nn_cfg_readl(hw, NFP_NET_CFG_STS);
+
+   memset(&link, 0, sizeof(struct rte_eth_link));
+
+   if (nn_link_status & NFP_NET_CFG_STS_LINK)
+   link.link_status = 1;
+
+   link.link_duplex = ETH_LINK_FULL_DUPLEX;
+   /* Other cards can limit the tx and rx rate per VF */
+   link.link_speed = ETH_LINK_SPEED_40G;
+
+   if (old.link_status != link.link_status) {
+   nfp_net_dev_atomic_write_link_status(dev, &link);
+   if (link.link_status)
+   PMD_DRV_LOG(INFO, "NIC Link is Up\n");
+   else
+   PMD_DRV_LOG(INFO, "NIC Link is Down\n");
+   return 0;
+   }
+
+   return -1;
+}
+
 static void
 nfp_net_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
 {
@@ -1895,6 +1990,7 @@ static struct eth_dev_ops nfp_net_eth_dev_ops = {
.dev_start  = nfp_net_start,
.dev_stop   = nfp_net_stop,
.dev_close  = nfp_net_close,
+   .link_update= nfp_net_link_update,
.stats_get  = nfp_net_stats_get,
.stats_reset= nfp_net_stats_reset,
.reta_update= nfp_net_reta_update,
-- 
1.7.9.5

[dpdk-dev] [PATCH v7 2/9] nfp: adding rx/tx functionality

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  993 +
 1 file changed, 993 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index 9938f5d..fe62b4e 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -74,8 +74,25 @@
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
 static int nfp_net_init(struct rte_eth_dev *eth_dev);
+static int nfp_net_rx_fill_freelist(struct nfp_net_rxq *rxq);
+static uint32_t nfp_net_rx_queue_count(struct rte_eth_dev *dev,
+  uint16_t queue_idx);
+static uint16_t nfp_net_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts);
+static void nfp_net_rx_queue_release(void *rxq);
+static int nfp_net_rx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
+ uint16_t nb_desc, unsigned int socket_id,
+ const struct rte_eth_rxconf *rx_conf,
+ struct rte_mempool *mp);
+static int nfp_net_tx_free_bufs(struct nfp_net_txq *txq);
+static void nfp_net_tx_queue_release(void *txq);
+static int nfp_net_tx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
+ uint16_t nb_desc, unsigned int socket_id,
+ const struct rte_eth_txconf *tx_conf);
 static int nfp_net_start(struct rte_eth_dev *dev);
 static void nfp_net_stop(struct rte_eth_dev *dev);
+static uint16_t nfp_net_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
+ uint16_t nb_pkts);

 /*
  * The offset of the queue controller queues in the PCIe Target. These
@@ -186,6 +203,100 @@ nn_cfg_writeq(struct nfp_net_hw *hw, int off, uint64_t 
val)
nn_writeq(rte_cpu_to_le_64(val), hw->ctrl_bar + off);
 }

+/* Creating memzone for hardware rings. */
+static const struct rte_memzone *
+ring_dma_zone_reserve(struct rte_eth_dev *dev, const char *ring_name,
+ uint16_t queue_id, uint32_t ring_size, int socket_id)
+{
+   char z_name[RTE_MEMZONE_NAMESIZE];
+   const struct rte_memzone *mz;
+
+   snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
+dev->driver->pci_drv.name,
+ring_name, dev->data->port_id, queue_id);
+
+   mz = rte_memzone_lookup(z_name);
+   if (mz)
+   return mz;
+
+   return rte_memzone_reserve_aligned(z_name, ring_size, socket_id, 0,
+  NFP_MEMZONE_ALIGN);
+}
+
+static void
+nfp_net_rx_queue_release_mbufs(struct nfp_net_rxq *rxq)
+{
+   unsigned i;
+
+   if (rxq->rxbufs == NULL)
+   return;
+
+   for (i = 0; i < rxq->rx_count; i++) {
+   if (rxq->rxbufs[i].mbuf) {
+   rte_pktmbuf_free_seg(rxq->rxbufs[i].mbuf);
+   rxq->rxbufs[i].mbuf = NULL;
+   }
+   }
+}
+
+static void
+nfp_net_rx_queue_release(void *rx_queue)
+{
+   struct nfp_net_rxq *rxq = rx_queue;
+
+   if (rxq) {
+   nfp_net_rx_queue_release_mbufs(rxq);
+   rte_free(rxq->rxbufs);
+   rte_free(rxq);
+   }
+}
+
+static void
+nfp_net_reset_rx_queue(struct nfp_net_rxq *rxq)
+{
+   nfp_net_rx_queue_release_mbufs(rxq);
+   rxq->wr_p = 0;
+   rxq->rd_p = 0;
+   rxq->nb_rx_hold = 0;
+}
+
+static void
+nfp_net_tx_queue_release_mbufs(struct nfp_net_txq *txq)
+{
+   unsigned i;
+
+   if (txq->txbufs == NULL)
+   return;
+
+   for (i = 0; i < txq->tx_count; i++) {
+   if (txq->txbufs[i].mbuf) {
+   rte_pktmbuf_free_seg(txq->txbufs[i].mbuf);
+   txq->txbufs[i].mbuf = NULL;
+   }
+   }
+}
+
+static void
+nfp_net_tx_queue_release(void *tx_queue)
+{
+   struct nfp_net_txq *txq = tx_queue;
+
+   if (txq) {
+   nfp_net_tx_queue_release_mbufs(txq);
+   rte_free(txq->txbufs);
+   rte_free(txq);
+   }
+}
+
+static void
+nfp_net_reset_tx_queue(struct nfp_net_txq *txq)
+{
+   nfp_net_tx_queue_release_mbufs(txq);
+   txq->wr_p = 0;
+   txq->rd_p = 0;
+   txq->tail = 0;
+}
+
 static int
 __nfp_net_reconfig(struct nfp_net_hw *hw, uint32_t update)
 {
@@ -423,6 +534,18 @@ nfp_net_disable_queues(struct rte_eth_dev *dev)
hw->ctrl = new_ctrl;
 }

+static int
+nfp_net_rx_freelist_setup(struct rte_eth_dev *dev)
+{
+   int i;
+
+   for (i = 0; i < dev->data->nb_rx_queues; i++) {
+   if (nfp_net_rx_fill_freelist(dev->data->rx_queues[i]) < 0)
+   return -1;
+   }
+   return 0;
+}
+
 static void
 nfp_net_params_setup(struct nfp_net_hw *hw)
 {
@@ -451,6 +574,7 @@ nfp_net_start(struct rte_eth_dev *dev)
 {
uint32_t

[dpdk-dev] [PATCH v7 7/9] nfp: link status change interrupt support

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  123 +
 1 file changed, 123 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index b099091..d99b0cc 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -73,6 +73,9 @@
 /* Prototypes */
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
+static void nfp_net_dev_interrupt_handler(struct rte_intr_handle *handle,
+ void *param);
+static void nfp_net_dev_interrupt_delayed_handler(void *param);
 static int nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
 static void nfp_net_infos_get(struct rte_eth_dev *dev,
  struct rte_eth_dev_info *dev_info);
@@ -731,6 +734,7 @@ nfp_net_close(struct rte_eth_dev *dev)

nfp_net_stop(dev);

+   rte_intr_disable(&dev->pci_dev->intr_handle);
nn_cfg_writeb(hw, NFP_NET_CFG_LSC, 0xff);

/*
@@ -1115,6 +1119,114 @@ nfp_net_rx_queue_count(struct rte_eth_dev *dev, 
uint16_t queue_idx)
return count;
 }

+static void
+nfp_net_dev_link_status_print(struct rte_eth_dev *dev)
+{
+   struct rte_eth_link link;
+
+   memset(&link, 0, sizeof(link));
+   nfp_net_dev_atomic_read_link_status(dev, &link);
+   if (link.link_status)
+   RTE_LOG(INFO, PMD, "Port %d: Link Up - speed %u Mbps - %s\n",
+   (int)(dev->data->port_id), (unsigned)link.link_speed,
+   link.link_duplex == ETH_LINK_FULL_DUPLEX
+   ? "full-duplex" : "half-duplex");
+   else
+   RTE_LOG(INFO, PMD, " Port %d: Link Down\n",
+   (int)(dev->data->port_id));
+
+   RTE_LOG(INFO, PMD, "PCI Address: %04d:%02d:%02d:%d\n",
+   dev->pci_dev->addr.domain, dev->pci_dev->addr.bus,
+   dev->pci_dev->addr.devid, dev->pci_dev->addr.function);
+}
+
+/* Interrupt configuration and handling */
+
+/*
+ * nfp_net_irq_unmask - Unmask an interrupt
+ *
+ * If MSI-X auto-masking is enabled clear the mask bit, otherwise
+ * clear the ICR for the entry.
+ */
+static void
+nfp_net_irq_unmask(struct rte_eth_dev *dev)
+{
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (hw->ctrl & NFP_NET_CFG_CTRL_MSIXAUTO) {
+   /* If MSI-X auto-masking is used, clear the entry */
+   rte_wmb();
+   rte_intr_enable(&dev->pci_dev->intr_handle);
+   } else {
+   /* Make sure all updates are written before un-masking */
+   rte_wmb();
+   nn_cfg_writeb(hw, NFP_NET_CFG_ICR(NFP_NET_IRQ_LSC_IDX),
+ NFP_NET_CFG_ICR_UNMASKED);
+   }
+}
+
+static void
+nfp_net_dev_interrupt_handler(__rte_unused struct rte_intr_handle *handle,
+ void *param)
+{
+   int64_t timeout;
+   struct rte_eth_link link;
+   struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+   PMD_DRV_LOG(DEBUG, "We got a LSC interrupt!!!\n");
+
+   /* get the link status */
+   memset(&link, 0, sizeof(link));
+   nfp_net_dev_atomic_read_link_status(dev, &link);
+
+   nfp_net_link_update(dev, 0);
+
+   /* likely to up */
+   if (!link.link_status) {
+   /* handle it 1 sec later, wait it being stable */
+   timeout = NFP_NET_LINK_UP_CHECK_TIMEOUT;
+   /* likely to down */
+   } else {
+   /* handle it 4 sec later, wait it being stable */
+   timeout = NFP_NET_LINK_DOWN_CHECK_TIMEOUT;
+   }
+
+   if (rte_eal_alarm_set(timeout * 1000,
+ nfp_net_dev_interrupt_delayed_handler,
+ (void *)dev) < 0) {
+   RTE_LOG(ERR, PMD, "Error setting alarm");
+   /* Unmasking */
+   nfp_net_irq_unmask(dev);
+   }
+}
+
+/*
+ * Interrupt handler which shall be registered for alarm callback for delayed
+ * handling specific interrupt to wait for the stable nic state. As the NIC
+ * interrupt state is not stable for nfp after link is just down, it needs
+ * to wait 4 seconds to get the stable status.
+ *
+ * @param handle   Pointer to interrupt handle.
+ * @param paramThe address of parameter (struct rte_eth_dev *)
+ *
+ * @return  void
+ */
+static void
+nfp_net_dev_interrupt_delayed_handler(void *param)
+{
+   struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+   nfp_net_link_update(dev, 0);
+   _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC);
+
+   nfp_net_dev_link_status_print(dev);
+
+   /* Unmasking */
+   nfp_net_irq_unmask(dev);
+}
+
 static int
 nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 {
@@ -2315,6 +2427,17 @@ nfp_net_init(struct rte_eth_de

[dpdk-dev] [PATCH v7 6/9] nfp: adding extra functionality

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  191 +
 1 file changed, 191 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index a7db4bc..b099091 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -73,8 +73,13 @@
 /* Prototypes */
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
+static int nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
+static void nfp_net_infos_get(struct rte_eth_dev *dev,
+ struct rte_eth_dev_info *dev_info);
 static int nfp_net_init(struct rte_eth_dev *eth_dev);
 static int nfp_net_link_update(struct rte_eth_dev *dev, int wait_to_complete);
+static void nfp_net_promisc_enable(struct rte_eth_dev *dev);
+static void nfp_net_promisc_disable(struct rte_eth_dev *dev);
 static int nfp_net_rx_fill_freelist(struct nfp_net_rxq *rxq);
 static uint32_t nfp_net_rx_queue_count(struct rte_eth_dev *dev,
   uint16_t queue_idx);
@@ -734,6 +739,65 @@ nfp_net_close(struct rte_eth_dev *dev)
 */
 }

+static void
+nfp_net_promisc_enable(struct rte_eth_dev *dev)
+{
+   uint32_t new_ctrl, update = 0;
+   struct nfp_net_hw *hw;
+
+   PMD_DRV_LOG(DEBUG, "Promiscuous mode enable\n");
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (!(hw->cap & NFP_NET_CFG_CTRL_PROMISC)) {
+   PMD_INIT_LOG(INFO, "Promiscuous mode not supported\n");
+   return;
+   }
+
+   if (hw->ctrl & NFP_NET_CFG_CTRL_PROMISC) {
+   PMD_DRV_LOG(INFO, "Promiscuous mode already enabled\n");
+   return;
+   }
+
+   new_ctrl = hw->ctrl | NFP_NET_CFG_CTRL_PROMISC;
+   update = NFP_NET_CFG_UPDATE_GEN;
+
+   /*
+* DPDK sets promiscuous mode on just after this call assuming
+* it can not fail ...
+*/
+   if (nfp_net_reconfig(hw, new_ctrl, update) < 0)
+   return;
+
+   hw->ctrl = new_ctrl;
+}
+
+static void
+nfp_net_promisc_disable(struct rte_eth_dev *dev)
+{
+   uint32_t new_ctrl, update = 0;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if ((hw->ctrl & NFP_NET_CFG_CTRL_PROMISC) == 0) {
+   PMD_DRV_LOG(INFO, "Promiscuous mode already disabled\n");
+   return;
+   }
+
+   new_ctrl = hw->ctrl & ~NFP_NET_CFG_CTRL_PROMISC;
+   update = NFP_NET_CFG_UPDATE_GEN;
+
+   /*
+* DPDK sets promiscuous mode off just before this call
+* assuming it can not fail ...
+*/
+   if (nfp_net_reconfig(hw, new_ctrl, update) < 0)
+   return;
+
+   hw->ctrl = new_ctrl;
+}
+
 /*
  * return 0 means link status changed, -1 means not changed
  *
@@ -948,6 +1012,65 @@ nfp_net_stats_reset(struct rte_eth_dev *dev)
nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_DISCARDS);
 }

+static void
+nfp_net_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
+{
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   dev_info->driver_name = dev->driver->pci_drv.name;
+   dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
+   dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
+   dev_info->min_rx_bufsize = ETHER_MIN_MTU;
+   dev_info->max_rx_pktlen = hw->mtu;
+   /* Next should change when PF support is implemented */
+   dev_info->max_mac_addrs = 1;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_RXVLAN)
+   dev_info->rx_offload_capa = DEV_RX_OFFLOAD_VLAN_STRIP;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_RXCSUM)
+   dev_info->rx_offload_capa |= DEV_RX_OFFLOAD_IPV4_CKSUM |
+DEV_RX_OFFLOAD_UDP_CKSUM |
+DEV_RX_OFFLOAD_TCP_CKSUM;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_TXVLAN)
+   dev_info->tx_offload_capa = DEV_TX_OFFLOAD_VLAN_INSERT;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_TXCSUM)
+   dev_info->tx_offload_capa |= DEV_TX_OFFLOAD_IPV4_CKSUM |
+DEV_RX_OFFLOAD_UDP_CKSUM |
+DEV_RX_OFFLOAD_TCP_CKSUM;
+
+   dev_info->default_rxconf = (struct rte_eth_rxconf) {
+   .rx_thresh = {
+   .pthresh = DEFAULT_RX_PTHRESH,
+   .hthresh = DEFAULT_RX_HTHRESH,
+   .wthresh = DEFAULT_RX_WTHRESH,
+   },
+   .rx_free_thresh = DEFAULT_RX_FREE_THRESH,
+   .rx_drop_en = 0,
+   };
+
+   dev_info->default_txconf = (struct rte_eth_txconf) {
+   .tx_thresh = {
+   .pthresh = DEFAULT_TX_PTHRESH,
+   .hthresh = DEFAULT_TX_

[dpdk-dev] [PATCH v7 9/9] nfp: Updating MAINTAINERS

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 MAINTAINERS |1 +
 1 file changed, 1 insertion(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index df5b962..b265893 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -321,6 +321,7 @@ F: drivers/net/null/
 Netronome nfp
 M: Alejandro Lucero 
 F: drivers/net/nfp/
+F: doc/guides/nics/nfp.rst

 Packet processing
 -
-- 
1.7.9.5

[dpdk-dev] [PATCH v7 1/9] nfp: basic initialization

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 MAINTAINERS  |3 +
 config/common_linuxapp   |6 +
 doc/guides/rel_notes/release_2_2.rst |4 +
 drivers/net/Makefile |1 +
 drivers/net/nfp/Makefile |   57 +++
 drivers/net/nfp/nfp_net.c|  699 ++
 drivers/net/nfp/nfp_net_ctrl.h   |  324 
 drivers/net/nfp/nfp_net_logs.h   |   75 
 drivers/net/nfp/nfp_net_pmd.h|  453 ++
 mk/rte.app.mk|1 +
 10 files changed, 1623 insertions(+)
 create mode 100644 drivers/net/nfp/Makefile
 create mode 100644 drivers/net/nfp/nfp_net.c
 create mode 100644 drivers/net/nfp/nfp_net_ctrl.h
 create mode 100644 drivers/net/nfp/nfp_net_logs.h
 create mode 100644 drivers/net/nfp/nfp_net_pmd.h

diff --git a/MAINTAINERS b/MAINTAINERS
index 840faeb..df5b962 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -318,6 +318,9 @@ Null PMD
 M: Tetsuya Mukawa 
 F: drivers/net/null/

+Netronome nfp
+M: Alejandro Lucero 
+F: drivers/net/nfp/

 Packet processing
 -
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 4a68da4..1d77db7 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -279,6 +279,12 @@ CONFIG_RTE_LIBRTE_VMXNET3_DEBUG_TX_FREE=n
 CONFIG_RTE_LIBRTE_VMXNET3_DEBUG_DRIVER=n

 #
+# Compile burst-oriented Netronome NFP PMD driver
+#
+CONFIG_RTE_LIBRTE_NFP_PMD=n
+CONFIG_RTE_LIBRTE_NFP_DEBUG=n
+
+#
 # Compile example software rings based PMD
 #
 CONFIG_RTE_LIBRTE_PMD_RING=y
diff --git a/doc/guides/rel_notes/release_2_2.rst 
b/doc/guides/rel_notes/release_2_2.rst
index 8c77768..8154db7 100644
--- a/doc/guides/rel_notes/release_2_2.rst
+++ b/doc/guides/rel_notes/release_2_2.rst
@@ -195,6 +195,10 @@ Drivers

   Fixed issue when releasing null control queue.

+* **nfp: adding new PMD for Netronome nfp-6xxx card.**
+
+  Support for using Netronome nfp-6xxx with PCI VFs.
+

 Libraries
 ~
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index cddcd57..6e4497e 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -43,6 +43,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe
 DIRS-$(CONFIG_RTE_LIBRTE_MLX4_PMD) += mlx4
 DIRS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5
 DIRS-$(CONFIG_RTE_LIBRTE_MPIPE_PMD) += mpipe
+DIRS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += nfp
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_NULL) += null
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_PCAP) += pcap
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_RING) += ring
diff --git a/drivers/net/nfp/Makefile b/drivers/net/nfp/Makefile
new file mode 100644
index 000..a0c717e
--- /dev/null
+++ b/drivers/net/nfp/Makefile
@@ -0,0 +1,57 @@
+#   BSD LICENSE
+#
+#   Copyright(c) 2010-2014 Intel Corporation. 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 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 $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_pmd_nfp.a
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+OBJS_BASE_DRIVER=$(patsubst %.c,%.o,$(notdir $(wildcard 
$(RTE_SDK)/lib/librte_pmd_nfp/*.c)))
+$(foreach obj, $(OBJS_BASE_DRIVER), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
+
+VPATH += $(RTE_SDK)/drivers/net/nfp/
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += nfp_net.c
+
+# this lib depends upon:
+DEPDIRS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += lib/librte_eal lib/librte_ether
+DEPDIRS-$(CONF

[dpdk-dev] [PATCH v7 8/9] nfp: adding nic guide

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 doc/guides/nics/index.rst |1 +
 doc/guides/nics/nfp.rst   |  265 +
 2 files changed, 266 insertions(+)
 create mode 100644 doc/guides/nics/nfp.rst

diff --git a/doc/guides/nics/index.rst b/doc/guides/nics/index.rst
index 0a0b724..7bf2938 100644
--- a/doc/guides/nics/index.rst
+++ b/doc/guides/nics/index.rst
@@ -46,6 +46,7 @@ Network Interface Controller Drivers
 intel_vf
 mlx4
 mlx5
+nfp
 szedata2
 virtio
 vmxnet3
diff --git a/doc/guides/nics/nfp.rst b/doc/guides/nics/nfp.rst
new file mode 100644
index 000..06fe603
--- /dev/null
+++ b/doc/guides/nics/nfp.rst
@@ -0,0 +1,265 @@
+..  BSD LICENSE
+Copyright(c) 2015 Netronome Systems, Inc. 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 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.
+
+NFP poll mode driver library
+
+
+Netronome's sixth generation of flow processors pack 216 programmable
+cores and over 100 hardware accelerators that uniquely combine packet,
+flow, security and content processing in a single device that scales
+up to 400 Gbps.
+
+This document explains how to use DPDK with the Netronome Poll Mode
+Driver (PMD) supporting Netronome's Network Flow Processor 6xxx
+(NFP-6xxx).
+
+Currently the driver supports virtual functions (VFs) only.
+
+Dependencies
+
+
+Before using the Netronome's DPDK PMD some NFP-6xxx configuration,
+which is not related to DPDK, is required. The system requires
+installation of **Netronome's BSP (Board Support Package)** which includes
+Linux drivers, programs and libraries.
+
+If you have a NFP-6xxx device you should already have the code and
+documentation for doing this configuration. Contact
+**support at netronome.com** to obtain the latest available firmware.
+
+The NFP Linux kernel drivers (including the required PF driver for the
+NFP) are available on Github at
+**https://github.com/Netronome/nfp-drv-kmods** along with build
+instructions.
+
+DPDK runs in userspace and PMDs uses the Linux kernel UIO interface to
+allow access to physical devices from userspace. The NFP PMD requires
+a separate UIO driver, **nfp_uio**, to perform correct
+initialization. This driver is part of Netronome?s BSP and it is
+equivalent to Intel's igb_uio driver.
+
+Building the software
+-
+
+Netronome's PMD code is provided in the **drivers/net/nfp** directory.
+Because Netronome?s BSP dependencies the driver is disabled by default
+in DPDK build using **common_linuxapp configuration** file. Enabling the
+driver or if you use another configuration file and want to have NFP
+support, this variable is needed:
+
+- **CONFIG_RTE_LIBRTE_NFP_PMD=y**
+
+Once DPDK is built all the DPDK apps and examples include support for
+the NFP PMD.
+
+
+System configuration
+
+
+Using the NFP PMD is not different to using other PMDs. Usual steps are:
+
+#. **Configure hugepages:** All major Linux distributions have the hugepages
+   functionality enabled by default. By default this allows the system uses for
+   working with transparent hugepages. But in this case some hugepages need to
+   be created/reserved for use with the DPDK through the hugetlbfs file system.
+   First the virtual file system need to be mounted:
+
+   .. code-block:: console
+
+  mount -t hugetlbfs none /mnt/hugetlbfs
+
+   The command uses the common mount po

[dpdk-dev] [PATCH v7 1/9] nfp: basic initialization

[Thomas Monjalon]

2015-11-25 15:36, Alejandro.Lucero:
> +OBJS_BASE_DRIVER=$(patsubst %.c,%.o,$(notdir $(wildcard 
> $(RTE_SDK)/lib/librte_pmd_nfp/*.c)))
> +$(foreach obj, $(OBJS_BASE_DRIVER), $(eval 
> CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))

It seems you forgot to remove these lines.

[dpdk-dev] rte_eth_tx_queue_setup() failing (error -22) when setting up tx queues on a VMXNET3 port...

[Stephen Hemminger]

On Wed, 25 Nov 2015 10:32:33 +
"Montorsi, Francesco"  wrote:

> Hi all,
> I have a server running VMWare ESXi 5.5.0 and VMWare vCenter 5.5.0. On such 
> server I created a VM with two VMXNET3 NIC cards (one for management, the 
> other one should be used with DPDK to enable fast-RX of packets coming from 
> other VMs / bare-metal NICs).
> 
> Inside the VM I have successfully created 2MB hugepages, loaded igb_uio 
> driver and binded the 2nd VMXNET3 NIC to igb_uio driver:
> 
> --
> $ ./dpdk_nic_bind.py --status
> 
> Network devices using DPDK-compatible driver
> 
> :0b:00.0 'VMXNET3 Ethernet Controller' drv=igb_uio unused=vmxnet3
> 
> Network devices using kernel driver
> ===
> :03:00.0 'VMXNET3 Ethernet Controller' if=eth0 drv=vmxnet3 unused=igb_uio 
> *Active*
> 
> Other network devices
> =
> 
> --
> 
> However when I start my DPDK-application I get the following error during TX 
> queue initialization:
> 
> ERR rte_eth_tx_queue_setup: err=-22, port=0: Unknown error -22
> 
> This is the code that I'm using:
> 
>   ...
>rte_eth_dev_info_get(m_portid, m_dev_info);
> 
> 
>   // proceed with configuration
> 
>   struct rte_eth_conf port_conf;
>   memset(&port_conf, 0, sizeof(port_conf));
> 
>   port_conf.rxmode.mq_mode = ETH_MQ_RX_RSS;
>   /** The multi-queue packet distribution mode to be used, e.g. 
> RSS.; this is important to use multiple RX queues per port ID */
>   port_conf.rxmode.max_rx_pkt_len = ETHER_MAX_LEN;
>   port_conf.rxmode.hw_strip_crc   = 1; /**< enable CRC stripping by 
> hardware */
>   port_conf.rxmode.jumbo_frame= 1; /**< Jumbo Frame Support enabled */
>   //port_conf.rxmode.enable_lro = 1; /**< Enable LRO */   
> // NOT SUPPORTED ON TESTED HW
>   port_conf.rx_adv_conf.rss_conf.rss_hf = ETH_RSS_IP;
>   port_conf.txmode.mq_mode = ETH_MQ_TX_NONE;
> 
>   m_num_queues = MIN(64, m_dev_info->max_rx_queues);
>   m_num_queues = MIN(128, m_num_queues);
>   ret = rte_eth_dev_configure(m_portid, m_num_queues, 1 /* number of tx 
> queues */, &port_conf);
>   if (ret < 0)
>   {
>   HMLogError("HwEmulDPDKPort::init() rte_eth_dev_configure: 
> err=%d, port=%u: %s", ret, m_portid, rte_strerror(ret));
>   return false;
>   }
> 
>   // init one TX queue: even if we never ever TX packets, at least 1 
> queue is needed!
> --->  ret = rte_eth_tx_queue_setup(m_portid, 0 /* queue ID */, 64 /* num 
> descriptors */, rte_eth_dev_socket_id(m_portid), NULL);
>   if (ret < 0)
>   {
>   // retry with just 1 descriptor
> 
> --->  ret = rte_eth_tx_queue_setup(m_portid, 0 /* queue ID */, 1 /* 
> num descriptors */, rte_eth_dev_socket_id(m_portid), NULL);
>   if (ret < 0)
>   {
>   HMLogError("HwEmulDPDKPort::init() 
> rte_eth_tx_queue_setup: err=%d, port=%u: %s", ret, m_portid, 
> rte_strerror(ret));
>   return false;
>   }
>   }
> 
> 
> Basically since I want to only receive packets (no TX) I'm not really 
> interested in TX queues / num of TX descriptors... However I have troubles 
> decrypting this -22 error code... any hint?

First advice, DPDK is open source, therefore you have the source use it.

Error codes match Linux/Unix errno's. You can use strerror(-ret) to find the 
value
or look in errno.h. Hint -22 == EINVAL

If you read the source, you will see that it there are log messages enabled if 
you
configure with LIBRTE_ETHDEV_DEBUG enabled and rebuild DPDK. And there are log 
messages
for VMXNETE3 controlled by LIBRTE_VMXNET3_DEBUG_INIT

[dpdk-dev] [PATCH] mk: fix the combined library problems by replacing it with a linker script

[Stephen Hemminger]

On Wed, 25 Nov 2015 10:38:48 +0200
Panu Matilainen  wrote:

> On 11/25/2015 12:46 AM, Stephen Hemminger wrote:
> > On Tue, 24 Nov 2015 16:31:17 +0200
> > Panu Matilainen  wrote:
> >
> >> The physically linked-together combined library has been an increasing
> >> source of problems, as was predicted when library and symbol versioning
> >> was introduced. Replace the complex and fragile construction with a
> >> simple linker script which achieves the same without all the problems,
> >> remove the related kludges from eg mlx drivers.
> >>
> >> Since creating the linker script is practically zero cost, remove the
> >> config option and just create it always.
> >>
> >> Based on a patch by Sergio Gonzales Monroy, linker script approach
> >> initially suggested by Neil Horman.
> >>
> >> Suggested-by: Sergio Gonzalez Monroy 
> >> Suggested-by: Neil Horman 
> >> Signed-off-by: Panu Matilainen 
> >
> > But it now means distros have to ship 20 libraries which seems like
> > a step back.
> 
> That's how Fedora and RHEL are shipping it already and nobody has so 
> much as noticed anything strange, much less complained about it. 20 
> libraries is but a drop in the ocean on a average distro. But more to 
> the point, distros will prefer 50 working libraries over one that doesn't.
> 
> The combined library as it is simply is no longer a viable option. 
> Besides just being broken (witness the strange hacks people are coming 
> up with to work around issues in it) its ugly because it basically gives 
> the middle finger to all the effort going into version compatibility, 
> and its also big. Few projects will use every library in DPDK, but with 
> the combined library they're forced to lug the 800 pound gorilla along 
> needlessly.
> 
>   - Panu -
> 

Fixing the combined library took less than an hour for us.

[dpdk-dev] [PATCH v8 0/9] support for netronome nfp-6xxx card

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

This patchset adds a new PMD for Netronome nfp-6xxx card.
Just PCI Virtual Functions supported.
Using this PMD requires previous Netronome BSP installation.

v8:
 - removing remaining unnecessary flags to PMD Makefile

v7:
 - Adding support for link status changes interrupts.
 - removing unnecessary flags when compiling the PMD.

v6:
 - Making each patch compilable.

v5:
 - Splitting up patches per functionality.

v4:
 - Getting rid of nfp_uio. Just submitting PMD.

v3:
 - Making all patches independent for applying and building.
 - changing commits messages following standard.

v2:
 - Code style changes based on checkpatch.pl and DPDK style guide.
 - Documentation changes using the right rst format.
 - Moving the documentation files to a new patch file.
 - Adding info to MAINTAINERS and release files.

Alejandro.Lucero (9):
  nfp: basic initialization
  nfp: adding tx/tx functionality
  nfp: adding rss
  nfp: adding stats
  nfp: adding link functionality
  nfp: adding extra functionality
  nfp: link status change interrupt support
  nfp: adding nic guide
  nfp: Updating MAINTAINERS

 MAINTAINERS  |4 +
 config/common_linuxapp   |6 +
 doc/guides/nics/index.rst|1 +
 doc/guides/nics/nfp.rst  |  265 
 doc/guides/rel_notes/release_2_2.rst |4 +
 drivers/net/Makefile |1 +
 drivers/net/nfp/Makefile |   54 +
 drivers/net/nfp/nfp_net.c| 2499 ++
 drivers/net/nfp/nfp_net_ctrl.h   |  324 +
 drivers/net/nfp/nfp_net_logs.h   |   75 +
 drivers/net/nfp/nfp_net_pmd.h|  453 ++
 mk/rte.app.mk|1 +
 12 files changed, 3687 insertions(+)
 create mode 100644 doc/guides/nics/nfp.rst
 create mode 100644 drivers/net/nfp/Makefile
 create mode 100644 drivers/net/nfp/nfp_net.c
 create mode 100644 drivers/net/nfp/nfp_net_ctrl.h
 create mode 100644 drivers/net/nfp/nfp_net_logs.h
 create mode 100644 drivers/net/nfp/nfp_net_pmd.h

-- 
1.7.9.5

[dpdk-dev] [PATCH v8 3/9] nfp: adding rss

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  218 +
 1 file changed, 218 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index fe62b4e..e40b613 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -1501,12 +1501,230 @@ xmit_end:
return i;
 }

+/* Update Redirection Table(RETA) of Receive Side Scaling of Ethernet device */
+static int
+nfp_net_reta_update(struct rte_eth_dev *dev,
+   struct rte_eth_rss_reta_entry64 *reta_conf,
+   uint16_t reta_size)
+{
+   uint32_t reta, mask;
+   int i, j;
+   int idx, shift;
+   uint32_t update;
+   struct nfp_net_hw *hw =
+   NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (!(hw->ctrl & NFP_NET_CFG_CTRL_RSS))
+   return -EINVAL;
+
+   if (reta_size != NFP_NET_CFG_RSS_ITBL_SZ) {
+   RTE_LOG(ERR, PMD, "The size of hash lookup table configured "
+   "(%d) doesn't match the number hardware can supported "
+   "(%d)\n", reta_size, NFP_NET_CFG_RSS_ITBL_SZ);
+   return -EINVAL;
+   }
+
+   /*
+* Update Redirection Table. There are 128 8bit-entries which can be
+* manage as 32 32bit-entries
+*/
+   for (i = 0; i < reta_size; i += 4) {
+   /* Handling 4 RSS entries per loop */
+   idx = i / RTE_RETA_GROUP_SIZE;
+   shift = i % RTE_RETA_GROUP_SIZE;
+   mask = (uint8_t)((reta_conf[idx].mask >> shift) & 0xF);
+
+   if (!mask)
+   continue;
+
+   reta = 0;
+   /* If all 4 entries were set, don't need read RETA register */
+   if (mask != 0xF)
+   reta = nn_cfg_readl(hw, NFP_NET_CFG_RSS_ITBL + i);
+
+   for (j = 0; j < 4; j++) {
+   if (!(mask & (0x1 << j)))
+   continue;
+   if (mask != 0xF)
+   /* Clearing the entry bits */
+   reta &= ~(0xFF << (8 * j));
+   reta |= reta_conf[idx].reta[shift + j] << (8 * j);
+   }
+   nn_cfg_writel(hw, NFP_NET_CFG_RSS_ITBL + shift, reta);
+   }
+
+   update = NFP_NET_CFG_UPDATE_RSS;
+
+   if (nfp_net_reconfig(hw, hw->ctrl, update) < 0)
+   return -EIO;
+
+   return 0;
+}
+
+ /* Query Redirection Table(RETA) of Receive Side Scaling of Ethernet device. 
*/
+static int
+nfp_net_reta_query(struct rte_eth_dev *dev,
+  struct rte_eth_rss_reta_entry64 *reta_conf,
+  uint16_t reta_size)
+{
+   uint8_t i, j, mask;
+   int idx, shift;
+   uint32_t reta;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (!(hw->ctrl & NFP_NET_CFG_CTRL_RSS))
+   return -EINVAL;
+
+   if (reta_size != NFP_NET_CFG_RSS_ITBL_SZ) {
+   RTE_LOG(ERR, PMD, "The size of hash lookup table configured "
+   "(%d) doesn't match the number hardware can supported "
+   "(%d)\n", reta_size, NFP_NET_CFG_RSS_ITBL_SZ);
+   return -EINVAL;
+   }
+
+   /*
+* Reading Redirection Table. There are 128 8bit-entries which can be
+* manage as 32 32bit-entries
+*/
+   for (i = 0; i < reta_size; i += 4) {
+   /* Handling 4 RSS entries per loop */
+   idx = i / RTE_RETA_GROUP_SIZE;
+   shift = i % RTE_RETA_GROUP_SIZE;
+   mask = (uint8_t)((reta_conf[idx].mask >> shift) & 0xF);
+
+   if (!mask)
+   continue;
+
+   reta = nn_cfg_readl(hw, NFP_NET_CFG_RSS_ITBL + shift);
+   for (j = 0; j < 4; j++) {
+   if (!(mask & (0x1 << j)))
+   continue;
+   reta_conf->reta[shift + j] =
+   (uint8_t)((reta >> (8 * j)) & 0xF);
+   }
+   }
+   return 0;
+}
+
+static int
+nfp_net_rss_hash_update(struct rte_eth_dev *dev,
+   struct rte_eth_rss_conf *rss_conf)
+{
+   uint32_t update;
+   uint32_t cfg_rss_ctrl = 0;
+   uint8_t key;
+   uint64_t rss_hf;
+   int i;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   rss_hf = rss_conf->rss_hf;
+
+   /* Checking if RSS is enabled */
+   if (!(hw->ctrl & NFP_NET_CFG_CTRL_RSS)) {
+   if (rss_hf != 0) { /* Enable RSS? */
+   RTE_LOG(ERR, PMD, "RSS unsupported\n");
+   return -EINVAL;
+   }
+   return 0; /* Nothing to do */
+   }
+
+   if (rss_conf->rss_key

[dpdk-dev] [PATCH v8 4/9] nfp: adding stats

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  179 +
 1 file changed, 179 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index e40b613..49c71bf 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -90,6 +90,9 @@ static int nfp_net_tx_queue_setup(struct rte_eth_dev *dev, 
uint16_t queue_idx,
  uint16_t nb_desc, unsigned int socket_id,
  const struct rte_eth_txconf *tx_conf);
 static int nfp_net_start(struct rte_eth_dev *dev);
+static void nfp_net_stats_get(struct rte_eth_dev *dev,
+ struct rte_eth_stats *stats);
+static void nfp_net_stats_reset(struct rte_eth_dev *dev);
 static void nfp_net_stop(struct rte_eth_dev *dev);
 static uint16_t nfp_net_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
  uint16_t nb_pkts);
@@ -679,6 +682,177 @@ nfp_net_close(struct rte_eth_dev *dev)
 */
 }

+static void
+nfp_net_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
+{
+   int i;
+   struct nfp_net_hw *hw;
+   struct rte_eth_stats nfp_dev_stats;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   /* RTE_ETHDEV_QUEUE_STAT_CNTRS default value is 16 */
+
+   /* reading per RX ring stats */
+   for (i = 0; i < dev->data->nb_rx_queues; i++) {
+   if (i == RTE_ETHDEV_QUEUE_STAT_CNTRS)
+   break;
+
+   nfp_dev_stats.q_ipackets[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_RXR_STATS(i));
+
+   nfp_dev_stats.q_ipackets[i] -=
+   hw->eth_stats_base.q_ipackets[i];
+
+   nfp_dev_stats.q_ibytes[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_RXR_STATS(i) + 0x8);
+
+   nfp_dev_stats.q_ibytes[i] -=
+   hw->eth_stats_base.q_ibytes[i];
+   }
+
+   /* reading per TX ring stats */
+   for (i = 0; i < dev->data->nb_tx_queues; i++) {
+   if (i == RTE_ETHDEV_QUEUE_STAT_CNTRS)
+   break;
+
+   nfp_dev_stats.q_opackets[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_TXR_STATS(i));
+
+   nfp_dev_stats.q_opackets[i] -=
+   hw->eth_stats_base.q_opackets[i];
+
+   nfp_dev_stats.q_obytes[i] =
+   nn_cfg_readq(hw, NFP_NET_CFG_TXR_STATS(i) + 0x8);
+
+   nfp_dev_stats.q_obytes[i] -=
+   hw->eth_stats_base.q_obytes[i];
+   }
+
+   nfp_dev_stats.ipackets =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_FRAMES);
+
+   nfp_dev_stats.ipackets -= hw->eth_stats_base.ipackets;
+
+   nfp_dev_stats.ibytes =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_OCTETS);
+
+   nfp_dev_stats.ibytes -= hw->eth_stats_base.ibytes;
+
+   nfp_dev_stats.opackets =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_TX_FRAMES);
+
+   nfp_dev_stats.opackets -= hw->eth_stats_base.opackets;
+
+   nfp_dev_stats.obytes =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_TX_OCTETS);
+
+   nfp_dev_stats.obytes -= hw->eth_stats_base.obytes;
+
+   nfp_dev_stats.imcasts =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_MC_FRAMES);
+
+   nfp_dev_stats.imcasts -= hw->eth_stats_base.imcasts;
+
+   /* reading general device stats */
+   nfp_dev_stats.ierrors =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_ERRORS);
+
+   nfp_dev_stats.ierrors -= hw->eth_stats_base.ierrors;
+
+   nfp_dev_stats.oerrors =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_TX_ERRORS);
+
+   nfp_dev_stats.oerrors -= hw->eth_stats_base.oerrors;
+
+   /* Multicast frames received */
+   nfp_dev_stats.imcasts =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_MC_FRAMES);
+
+   nfp_dev_stats.imcasts -= hw->eth_stats_base.imcasts;
+
+   /* RX ring mbuf allocation failures */
+   nfp_dev_stats.rx_nombuf = dev->data->rx_mbuf_alloc_failed;
+
+   nfp_dev_stats.imissed =
+   nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_DISCARDS);
+
+   nfp_dev_stats.imissed -= hw->eth_stats_base.imissed;
+
+   if (stats)
+   memcpy(stats, &nfp_dev_stats, sizeof(*stats));
+}
+
+static void
+nfp_net_stats_reset(struct rte_eth_dev *dev)
+{
+   int i;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   /*
+* hw->eth_stats_base records the per counter starting point.
+* Lets update it now
+*/
+
+   /* reading per RX ring stats */
+   for (i = 0; i < dev->data->nb_rx_queues; i++) {
+   if (i == RTE_ETHDEV_QUEUE_STAT_CNTRS)
+   break;
+
+   hw->eth_stats_base.q_ipackets[i] =
+   nn_cfg_rea

[dpdk-dev] [PATCH v8 2/9] nfp: adding tx/tx functionality

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  993 +
 1 file changed, 993 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index 9938f5d..fe62b4e 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -74,8 +74,25 @@
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
 static int nfp_net_init(struct rte_eth_dev *eth_dev);
+static int nfp_net_rx_fill_freelist(struct nfp_net_rxq *rxq);
+static uint32_t nfp_net_rx_queue_count(struct rte_eth_dev *dev,
+  uint16_t queue_idx);
+static uint16_t nfp_net_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
+ uint16_t nb_pkts);
+static void nfp_net_rx_queue_release(void *rxq);
+static int nfp_net_rx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
+ uint16_t nb_desc, unsigned int socket_id,
+ const struct rte_eth_rxconf *rx_conf,
+ struct rte_mempool *mp);
+static int nfp_net_tx_free_bufs(struct nfp_net_txq *txq);
+static void nfp_net_tx_queue_release(void *txq);
+static int nfp_net_tx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
+ uint16_t nb_desc, unsigned int socket_id,
+ const struct rte_eth_txconf *tx_conf);
 static int nfp_net_start(struct rte_eth_dev *dev);
 static void nfp_net_stop(struct rte_eth_dev *dev);
+static uint16_t nfp_net_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
+ uint16_t nb_pkts);

 /*
  * The offset of the queue controller queues in the PCIe Target. These
@@ -186,6 +203,100 @@ nn_cfg_writeq(struct nfp_net_hw *hw, int off, uint64_t 
val)
nn_writeq(rte_cpu_to_le_64(val), hw->ctrl_bar + off);
 }

+/* Creating memzone for hardware rings. */
+static const struct rte_memzone *
+ring_dma_zone_reserve(struct rte_eth_dev *dev, const char *ring_name,
+ uint16_t queue_id, uint32_t ring_size, int socket_id)
+{
+   char z_name[RTE_MEMZONE_NAMESIZE];
+   const struct rte_memzone *mz;
+
+   snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
+dev->driver->pci_drv.name,
+ring_name, dev->data->port_id, queue_id);
+
+   mz = rte_memzone_lookup(z_name);
+   if (mz)
+   return mz;
+
+   return rte_memzone_reserve_aligned(z_name, ring_size, socket_id, 0,
+  NFP_MEMZONE_ALIGN);
+}
+
+static void
+nfp_net_rx_queue_release_mbufs(struct nfp_net_rxq *rxq)
+{
+   unsigned i;
+
+   if (rxq->rxbufs == NULL)
+   return;
+
+   for (i = 0; i < rxq->rx_count; i++) {
+   if (rxq->rxbufs[i].mbuf) {
+   rte_pktmbuf_free_seg(rxq->rxbufs[i].mbuf);
+   rxq->rxbufs[i].mbuf = NULL;
+   }
+   }
+}
+
+static void
+nfp_net_rx_queue_release(void *rx_queue)
+{
+   struct nfp_net_rxq *rxq = rx_queue;
+
+   if (rxq) {
+   nfp_net_rx_queue_release_mbufs(rxq);
+   rte_free(rxq->rxbufs);
+   rte_free(rxq);
+   }
+}
+
+static void
+nfp_net_reset_rx_queue(struct nfp_net_rxq *rxq)
+{
+   nfp_net_rx_queue_release_mbufs(rxq);
+   rxq->wr_p = 0;
+   rxq->rd_p = 0;
+   rxq->nb_rx_hold = 0;
+}
+
+static void
+nfp_net_tx_queue_release_mbufs(struct nfp_net_txq *txq)
+{
+   unsigned i;
+
+   if (txq->txbufs == NULL)
+   return;
+
+   for (i = 0; i < txq->tx_count; i++) {
+   if (txq->txbufs[i].mbuf) {
+   rte_pktmbuf_free_seg(txq->txbufs[i].mbuf);
+   txq->txbufs[i].mbuf = NULL;
+   }
+   }
+}
+
+static void
+nfp_net_tx_queue_release(void *tx_queue)
+{
+   struct nfp_net_txq *txq = tx_queue;
+
+   if (txq) {
+   nfp_net_tx_queue_release_mbufs(txq);
+   rte_free(txq->txbufs);
+   rte_free(txq);
+   }
+}
+
+static void
+nfp_net_reset_tx_queue(struct nfp_net_txq *txq)
+{
+   nfp_net_tx_queue_release_mbufs(txq);
+   txq->wr_p = 0;
+   txq->rd_p = 0;
+   txq->tail = 0;
+}
+
 static int
 __nfp_net_reconfig(struct nfp_net_hw *hw, uint32_t update)
 {
@@ -423,6 +534,18 @@ nfp_net_disable_queues(struct rte_eth_dev *dev)
hw->ctrl = new_ctrl;
 }

+static int
+nfp_net_rx_freelist_setup(struct rte_eth_dev *dev)
+{
+   int i;
+
+   for (i = 0; i < dev->data->nb_rx_queues; i++) {
+   if (nfp_net_rx_fill_freelist(dev->data->rx_queues[i]) < 0)
+   return -1;
+   }
+   return 0;
+}
+
 static void
 nfp_net_params_setup(struct nfp_net_hw *hw)
 {
@@ -451,6 +574,7 @@ nfp_net_start(struct rte_eth_dev *dev)
 {
uint32_t

[dpdk-dev] [PATCH v8 5/9] nfp: adding link functionality

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |   96 +
 1 file changed, 96 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index 49c71bf..a7db4bc 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -74,6 +74,7 @@
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
 static int nfp_net_init(struct rte_eth_dev *eth_dev);
+static int nfp_net_link_update(struct rte_eth_dev *dev, int wait_to_complete);
 static int nfp_net_rx_fill_freelist(struct nfp_net_rxq *rxq);
 static uint32_t nfp_net_rx_queue_count(struct rte_eth_dev *dev,
   uint16_t queue_idx);
@@ -226,6 +227,57 @@ ring_dma_zone_reserve(struct rte_eth_dev *dev, const char 
*ring_name,
   NFP_MEMZONE_ALIGN);
 }

+/*
+ * Atomically reads 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
+nfp_net_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
+nfp_net_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
 nfp_net_rx_queue_release_mbufs(struct nfp_net_rxq *rxq)
 {
@@ -682,6 +734,49 @@ nfp_net_close(struct rte_eth_dev *dev)
 */
 }

+/*
+ * return 0 means link status changed, -1 means not changed
+ *
+ * Wait to complete is needed as it can take up to 9 seconds to get the Link
+ * status.
+ */
+static int
+nfp_net_link_update(struct rte_eth_dev *dev, __rte_unused int wait_to_complete)
+{
+   struct nfp_net_hw *hw;
+   struct rte_eth_link link, old;
+   uint32_t nn_link_status;
+
+   PMD_DRV_LOG(DEBUG, "Link update\n");
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   memset(&old, 0, sizeof(old));
+   nfp_net_dev_atomic_read_link_status(dev, &old);
+
+   nn_link_status = nn_cfg_readl(hw, NFP_NET_CFG_STS);
+
+   memset(&link, 0, sizeof(struct rte_eth_link));
+
+   if (nn_link_status & NFP_NET_CFG_STS_LINK)
+   link.link_status = 1;
+
+   link.link_duplex = ETH_LINK_FULL_DUPLEX;
+   /* Other cards can limit the tx and rx rate per VF */
+   link.link_speed = ETH_LINK_SPEED_40G;
+
+   if (old.link_status != link.link_status) {
+   nfp_net_dev_atomic_write_link_status(dev, &link);
+   if (link.link_status)
+   PMD_DRV_LOG(INFO, "NIC Link is Up\n");
+   else
+   PMD_DRV_LOG(INFO, "NIC Link is Down\n");
+   return 0;
+   }
+
+   return -1;
+}
+
 static void
 nfp_net_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
 {
@@ -1895,6 +1990,7 @@ static struct eth_dev_ops nfp_net_eth_dev_ops = {
.dev_start  = nfp_net_start,
.dev_stop   = nfp_net_stop,
.dev_close  = nfp_net_close,
+   .link_update= nfp_net_link_update,
.stats_get  = nfp_net_stats_get,
.stats_reset= nfp_net_stats_reset,
.reta_update= nfp_net_reta_update,
-- 
1.7.9.5

[dpdk-dev] [PATCH v8 9/9] nfp: Updating MAINTAINERS

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 MAINTAINERS |1 +
 1 file changed, 1 insertion(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index df5b962..ece737a 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -317,6 +317,7 @@ F: app/test/test_pmd_ring_perf.c
 Null PMD
 M: Tetsuya Mukawa 
 F: drivers/net/null/
+F: doc/guides/nics/nfp.rst

 Netronome nfp
 M: Alejandro Lucero 
-- 
1.7.9.5

[dpdk-dev] [PATCH v8 6/9] nfp: adding extra functionality

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  191 +
 1 file changed, 191 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index a7db4bc..b099091 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -73,8 +73,13 @@
 /* Prototypes */
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
+static int nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
+static void nfp_net_infos_get(struct rte_eth_dev *dev,
+ struct rte_eth_dev_info *dev_info);
 static int nfp_net_init(struct rte_eth_dev *eth_dev);
 static int nfp_net_link_update(struct rte_eth_dev *dev, int wait_to_complete);
+static void nfp_net_promisc_enable(struct rte_eth_dev *dev);
+static void nfp_net_promisc_disable(struct rte_eth_dev *dev);
 static int nfp_net_rx_fill_freelist(struct nfp_net_rxq *rxq);
 static uint32_t nfp_net_rx_queue_count(struct rte_eth_dev *dev,
   uint16_t queue_idx);
@@ -734,6 +739,65 @@ nfp_net_close(struct rte_eth_dev *dev)
 */
 }

+static void
+nfp_net_promisc_enable(struct rte_eth_dev *dev)
+{
+   uint32_t new_ctrl, update = 0;
+   struct nfp_net_hw *hw;
+
+   PMD_DRV_LOG(DEBUG, "Promiscuous mode enable\n");
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (!(hw->cap & NFP_NET_CFG_CTRL_PROMISC)) {
+   PMD_INIT_LOG(INFO, "Promiscuous mode not supported\n");
+   return;
+   }
+
+   if (hw->ctrl & NFP_NET_CFG_CTRL_PROMISC) {
+   PMD_DRV_LOG(INFO, "Promiscuous mode already enabled\n");
+   return;
+   }
+
+   new_ctrl = hw->ctrl | NFP_NET_CFG_CTRL_PROMISC;
+   update = NFP_NET_CFG_UPDATE_GEN;
+
+   /*
+* DPDK sets promiscuous mode on just after this call assuming
+* it can not fail ...
+*/
+   if (nfp_net_reconfig(hw, new_ctrl, update) < 0)
+   return;
+
+   hw->ctrl = new_ctrl;
+}
+
+static void
+nfp_net_promisc_disable(struct rte_eth_dev *dev)
+{
+   uint32_t new_ctrl, update = 0;
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if ((hw->ctrl & NFP_NET_CFG_CTRL_PROMISC) == 0) {
+   PMD_DRV_LOG(INFO, "Promiscuous mode already disabled\n");
+   return;
+   }
+
+   new_ctrl = hw->ctrl & ~NFP_NET_CFG_CTRL_PROMISC;
+   update = NFP_NET_CFG_UPDATE_GEN;
+
+   /*
+* DPDK sets promiscuous mode off just before this call
+* assuming it can not fail ...
+*/
+   if (nfp_net_reconfig(hw, new_ctrl, update) < 0)
+   return;
+
+   hw->ctrl = new_ctrl;
+}
+
 /*
  * return 0 means link status changed, -1 means not changed
  *
@@ -948,6 +1012,65 @@ nfp_net_stats_reset(struct rte_eth_dev *dev)
nn_cfg_readq(hw, NFP_NET_CFG_STATS_RX_DISCARDS);
 }

+static void
+nfp_net_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
+{
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   dev_info->driver_name = dev->driver->pci_drv.name;
+   dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
+   dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
+   dev_info->min_rx_bufsize = ETHER_MIN_MTU;
+   dev_info->max_rx_pktlen = hw->mtu;
+   /* Next should change when PF support is implemented */
+   dev_info->max_mac_addrs = 1;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_RXVLAN)
+   dev_info->rx_offload_capa = DEV_RX_OFFLOAD_VLAN_STRIP;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_RXCSUM)
+   dev_info->rx_offload_capa |= DEV_RX_OFFLOAD_IPV4_CKSUM |
+DEV_RX_OFFLOAD_UDP_CKSUM |
+DEV_RX_OFFLOAD_TCP_CKSUM;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_TXVLAN)
+   dev_info->tx_offload_capa = DEV_TX_OFFLOAD_VLAN_INSERT;
+
+   if (hw->cap & NFP_NET_CFG_CTRL_TXCSUM)
+   dev_info->tx_offload_capa |= DEV_TX_OFFLOAD_IPV4_CKSUM |
+DEV_RX_OFFLOAD_UDP_CKSUM |
+DEV_RX_OFFLOAD_TCP_CKSUM;
+
+   dev_info->default_rxconf = (struct rte_eth_rxconf) {
+   .rx_thresh = {
+   .pthresh = DEFAULT_RX_PTHRESH,
+   .hthresh = DEFAULT_RX_HTHRESH,
+   .wthresh = DEFAULT_RX_WTHRESH,
+   },
+   .rx_free_thresh = DEFAULT_RX_FREE_THRESH,
+   .rx_drop_en = 0,
+   };
+
+   dev_info->default_txconf = (struct rte_eth_txconf) {
+   .tx_thresh = {
+   .pthresh = DEFAULT_TX_PTHRESH,
+   .hthresh = DEFAULT_TX_

[dpdk-dev] [PATCH v8 7/9] nfp: link status change interrupt support

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 drivers/net/nfp/nfp_net.c |  123 +
 1 file changed, 123 insertions(+)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index b099091..d99b0cc 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -73,6 +73,9 @@
 /* Prototypes */
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
+static void nfp_net_dev_interrupt_handler(struct rte_intr_handle *handle,
+ void *param);
+static void nfp_net_dev_interrupt_delayed_handler(void *param);
 static int nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
 static void nfp_net_infos_get(struct rte_eth_dev *dev,
  struct rte_eth_dev_info *dev_info);
@@ -731,6 +734,7 @@ nfp_net_close(struct rte_eth_dev *dev)

nfp_net_stop(dev);

+   rte_intr_disable(&dev->pci_dev->intr_handle);
nn_cfg_writeb(hw, NFP_NET_CFG_LSC, 0xff);

/*
@@ -1115,6 +1119,114 @@ nfp_net_rx_queue_count(struct rte_eth_dev *dev, 
uint16_t queue_idx)
return count;
 }

+static void
+nfp_net_dev_link_status_print(struct rte_eth_dev *dev)
+{
+   struct rte_eth_link link;
+
+   memset(&link, 0, sizeof(link));
+   nfp_net_dev_atomic_read_link_status(dev, &link);
+   if (link.link_status)
+   RTE_LOG(INFO, PMD, "Port %d: Link Up - speed %u Mbps - %s\n",
+   (int)(dev->data->port_id), (unsigned)link.link_speed,
+   link.link_duplex == ETH_LINK_FULL_DUPLEX
+   ? "full-duplex" : "half-duplex");
+   else
+   RTE_LOG(INFO, PMD, " Port %d: Link Down\n",
+   (int)(dev->data->port_id));
+
+   RTE_LOG(INFO, PMD, "PCI Address: %04d:%02d:%02d:%d\n",
+   dev->pci_dev->addr.domain, dev->pci_dev->addr.bus,
+   dev->pci_dev->addr.devid, dev->pci_dev->addr.function);
+}
+
+/* Interrupt configuration and handling */
+
+/*
+ * nfp_net_irq_unmask - Unmask an interrupt
+ *
+ * If MSI-X auto-masking is enabled clear the mask bit, otherwise
+ * clear the ICR for the entry.
+ */
+static void
+nfp_net_irq_unmask(struct rte_eth_dev *dev)
+{
+   struct nfp_net_hw *hw;
+
+   hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (hw->ctrl & NFP_NET_CFG_CTRL_MSIXAUTO) {
+   /* If MSI-X auto-masking is used, clear the entry */
+   rte_wmb();
+   rte_intr_enable(&dev->pci_dev->intr_handle);
+   } else {
+   /* Make sure all updates are written before un-masking */
+   rte_wmb();
+   nn_cfg_writeb(hw, NFP_NET_CFG_ICR(NFP_NET_IRQ_LSC_IDX),
+ NFP_NET_CFG_ICR_UNMASKED);
+   }
+}
+
+static void
+nfp_net_dev_interrupt_handler(__rte_unused struct rte_intr_handle *handle,
+ void *param)
+{
+   int64_t timeout;
+   struct rte_eth_link link;
+   struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+   PMD_DRV_LOG(DEBUG, "We got a LSC interrupt!!!\n");
+
+   /* get the link status */
+   memset(&link, 0, sizeof(link));
+   nfp_net_dev_atomic_read_link_status(dev, &link);
+
+   nfp_net_link_update(dev, 0);
+
+   /* likely to up */
+   if (!link.link_status) {
+   /* handle it 1 sec later, wait it being stable */
+   timeout = NFP_NET_LINK_UP_CHECK_TIMEOUT;
+   /* likely to down */
+   } else {
+   /* handle it 4 sec later, wait it being stable */
+   timeout = NFP_NET_LINK_DOWN_CHECK_TIMEOUT;
+   }
+
+   if (rte_eal_alarm_set(timeout * 1000,
+ nfp_net_dev_interrupt_delayed_handler,
+ (void *)dev) < 0) {
+   RTE_LOG(ERR, PMD, "Error setting alarm");
+   /* Unmasking */
+   nfp_net_irq_unmask(dev);
+   }
+}
+
+/*
+ * Interrupt handler which shall be registered for alarm callback for delayed
+ * handling specific interrupt to wait for the stable nic state. As the NIC
+ * interrupt state is not stable for nfp after link is just down, it needs
+ * to wait 4 seconds to get the stable status.
+ *
+ * @param handle   Pointer to interrupt handle.
+ * @param paramThe address of parameter (struct rte_eth_dev *)
+ *
+ * @return  void
+ */
+static void
+nfp_net_dev_interrupt_delayed_handler(void *param)
+{
+   struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+   nfp_net_link_update(dev, 0);
+   _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC);
+
+   nfp_net_dev_link_status_print(dev);
+
+   /* Unmasking */
+   nfp_net_irq_unmask(dev);
+}
+
 static int
 nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 {
@@ -2315,6 +2427,17 @@ nfp_net_init(struct rte_eth_de

[dpdk-dev] [PATCH v8 8/9] nfp: adding nic guide

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 doc/guides/nics/index.rst |1 +
 doc/guides/nics/nfp.rst   |  265 +
 2 files changed, 266 insertions(+)
 create mode 100644 doc/guides/nics/nfp.rst

diff --git a/doc/guides/nics/index.rst b/doc/guides/nics/index.rst
index 0a0b724..7bf2938 100644
--- a/doc/guides/nics/index.rst
+++ b/doc/guides/nics/index.rst
@@ -46,6 +46,7 @@ Network Interface Controller Drivers
 intel_vf
 mlx4
 mlx5
+nfp
 szedata2
 virtio
 vmxnet3
diff --git a/doc/guides/nics/nfp.rst b/doc/guides/nics/nfp.rst
new file mode 100644
index 000..55ba64d
--- /dev/null
+++ b/doc/guides/nics/nfp.rst
@@ -0,0 +1,265 @@
+..  BSD LICENSE
+Copyright(c) 2015 Netronome Systems, Inc. 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 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.
+
+NFP poll mode driver library
+
+
+Netronome's sixth generation of flow processors pack 216 programmable
+cores and over 100 hardware accelerators that uniquely combine packet,
+flow, security and content processing in a single device that scales
+up to 400 Gbps.
+
+This document explains how to use DPDK with the Netronome Poll Mode
+Driver (PMD) supporting Netronome's Network Flow Processor 6xxx
+(NFP-6xxx).
+
+Currently the driver supports virtual functions (VFs) only.
+
+Dependencies
+
+
+Before using the Netronome's DPDK PMD some NFP-6xxx configuration,
+which is not related to DPDK, is required. The system requires
+installation of **Netronome's BSP (Board Support Package)** which includes
+Linux drivers, programs and libraries.
+
+If you have a NFP-6xxx device you should already have the code and
+documentation for doing this configuration. Contact
+**support at netronome.com** to obtain the latest available firmware.
+
+The NFP Linux kernel drivers (including the required PF driver for the
+NFP) are available on Github at
+**https://github.com/Netronome/nfp-drv-kmods** along with build
+instructions.
+
+DPDK runs in userspace and PMDs uses the Linux kernel UIO interface to
+allow access to physical devices from userspace. The NFP PMD requires
+a separate UIO driver, **nfp_uio**, to perform correct
+initialization. This driver is part of Netronome?s BSP and it is
+equivalent to Intel's igb_uio driver.
+
+Building the software
+-
+
+Netronome's PMD code is provided in the **drivers/net/nfp** directory.
+Because Netronome?s BSP dependencies the driver is disabled by default
+in DPDK build using **common_linuxapp configuration** file. Enabling the
+driver or if you use another configuration file and want to have NFP
+support, this variable is needed:
+
+- **CONFIG_RTE_LIBRTE_NFP_PMD=y**
+
+Once DPDK is built all the DPDK apps and examples include support for
+the NFP PMD.
+
+
+System configuration
+
+
+Using the NFP PMD is not different to using other PMDs. Usual steps are:
+
+#. **Configure hugepages:** All major Linux distributions have the hugepages
+   functionality enabled by default. By default this allows the system uses for
+   working with transparent hugepages. But in this case some hugepages need to
+   be created/reserved for use with the DPDK through the hugetlbfs file system.
+   First the virtual file system need to be mounted:
+
+   .. code-block:: console
+
+  mount -t hugetlbfs none /mnt/hugetlbfs
+
+   The command uses the common mount po

[dpdk-dev] [PATCH v8 1/9] nfp: basic initialization

[Alejandro.Lucero]

From: "Alejandro.Lucero" 

Signed-off-by: Alejandro.Lucero 
Signed-off-by: Rolf.Neugebauer 
---
 MAINTAINERS  |3 +
 config/common_linuxapp   |6 +
 doc/guides/rel_notes/release_2_2.rst |4 +
 drivers/net/Makefile |1 +
 drivers/net/nfp/Makefile |   54 +++
 drivers/net/nfp/nfp_net.c|  699 ++
 drivers/net/nfp/nfp_net_ctrl.h   |  324 
 drivers/net/nfp/nfp_net_logs.h   |   75 
 drivers/net/nfp/nfp_net_pmd.h|  453 ++
 mk/rte.app.mk|1 +
 10 files changed, 1620 insertions(+)
 create mode 100644 drivers/net/nfp/Makefile
 create mode 100644 drivers/net/nfp/nfp_net.c
 create mode 100644 drivers/net/nfp/nfp_net_ctrl.h
 create mode 100644 drivers/net/nfp/nfp_net_logs.h
 create mode 100644 drivers/net/nfp/nfp_net_pmd.h

diff --git a/MAINTAINERS b/MAINTAINERS
index 840faeb..df5b962 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -318,6 +318,9 @@ Null PMD
 M: Tetsuya Mukawa 
 F: drivers/net/null/

+Netronome nfp
+M: Alejandro Lucero 
+F: drivers/net/nfp/

 Packet processing
 -
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 4a68da4..1d77db7 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -279,6 +279,12 @@ CONFIG_RTE_LIBRTE_VMXNET3_DEBUG_TX_FREE=n
 CONFIG_RTE_LIBRTE_VMXNET3_DEBUG_DRIVER=n

 #
+# Compile burst-oriented Netronome NFP PMD driver
+#
+CONFIG_RTE_LIBRTE_NFP_PMD=n
+CONFIG_RTE_LIBRTE_NFP_DEBUG=n
+
+#
 # Compile example software rings based PMD
 #
 CONFIG_RTE_LIBRTE_PMD_RING=y
diff --git a/doc/guides/rel_notes/release_2_2.rst 
b/doc/guides/rel_notes/release_2_2.rst
index 8c77768..8154db7 100644
--- a/doc/guides/rel_notes/release_2_2.rst
+++ b/doc/guides/rel_notes/release_2_2.rst
@@ -195,6 +195,10 @@ Drivers

   Fixed issue when releasing null control queue.

+* **nfp: adding new PMD for Netronome nfp-6xxx card.**
+
+  Support for using Netronome nfp-6xxx with PCI VFs.
+

 Libraries
 ~
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index cddcd57..6e4497e 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -43,6 +43,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe
 DIRS-$(CONFIG_RTE_LIBRTE_MLX4_PMD) += mlx4
 DIRS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5
 DIRS-$(CONFIG_RTE_LIBRTE_MPIPE_PMD) += mpipe
+DIRS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += nfp
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_NULL) += null
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_PCAP) += pcap
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_RING) += ring
diff --git a/drivers/net/nfp/Makefile b/drivers/net/nfp/Makefile
new file mode 100644
index 000..06bdec8
--- /dev/null
+++ b/drivers/net/nfp/Makefile
@@ -0,0 +1,54 @@
+#   BSD LICENSE
+#
+#   Copyright(c) 2010-2014 Intel Corporation. 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 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 $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_pmd_nfp.a
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+VPATH += $(RTE_SDK)/drivers/net/nfp/
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += nfp_net.c
+
+# this lib depends upon:
+DEPDIRS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += lib/librte_eal lib/librte_ether
+DEPDIRS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += lib/librte_mempool lib/librte_mbuf
+DEPDIRS-$(CONFIG_RTE_LIBRTE_NFP_PMD) += lib/librte_net lib/librte_malloc
+
+include $(RTE_SDK)/mk/rte.lib.mk
diff --git

[dpdk-dev] [PATCH v8 5/9] nfp: adding link functionality

[Stephen Hemminger]

On Wed, 25 Nov 2015 16:19:51 +
"Alejandro.Lucero"  wrote:

> +/*
> + * Atomically reads 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
> +nfp_net_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;
> +}
> +
> +/

Sigh, this code has been copied and pasted to every driver.
Why is it not part of standard rte_ethdev code.

[dpdk-dev] [PATCH v8 5/9] nfp: adding link functionality

[Thomas Monjalon]

2015-11-25 08:29, Stephen Hemminger:
> On Wed, 25 Nov 2015 16:19:51 +
> "Alejandro.Lucero"  wrote:
> 
> > +/*
> > + * Atomically reads 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
> > +nfp_net_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;
> > +}
> > +
> > +/
> 
> Sigh, this code has been copied and pasted to every driver.
> Why is it not part of standard rte_ethdev code.

Because nobody made the patch.

Stephen, how is your mood today?

[dpdk-dev] [PATCH v8 7/9] nfp: link status change interrupt support

[Stephen Hemminger]

On Wed, 25 Nov 2015 16:19:53 +
"Alejandro.Lucero"  wrote:

> From: "Alejandro.Lucero" 
> 
> Signed-off-by: Alejandro.Lucero 
> Signed-off-by: Rolf.Neugebauer 
> ---

You need to set RTE_PCI_DRV_INTR_LSC in drv_flags of eth_driver to tell
upper layers that the driver supports link state interrutpt (or not).

[dpdk-dev] [PATCH v8 5/9] nfp: adding link functionality

[Alejandro Lucero]

I tried to do that but there is some issue with the inlining. I think this
is due to inline keyword being processed (also) as static by the compiler.

On Wed, Nov 25, 2015 at 4:29 PM, Stephen Hemminger <
stephen at networkplumber.org> wrote:

> On Wed, 25 Nov 2015 16:19:51 +
> "Alejandro.Lucero"  wrote:
>
> > +/*
> > + * Atomically reads 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
> > +nfp_net_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;
> > +}
> > +
> > +/
>
> Sigh, this code has been copied and pasted to every driver.
> Why is it not part of standard rte_ethdev code.
>

[dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict aliasing

[Mrzyglod, DanielX T]

If we change input buf:
xapp-gcc/include/rte_ip.h:211: error: dereferencing pointer 'u16' does break 
strict-aliasing rules
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:204: note: initialized 
from here
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:212: error: 
dereferencing pointer '({anonymous})' does break strict-aliasing rules
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:212: note: initialized 
from here
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:213: error: 
dereferencing pointer '({anonymous})' does break strict-aliasing rules
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:213: note: initialized 
from here
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:214: error: 
dereferencing pointer '({anonymous})' does break strict-aliasing rules
/root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:214: note: initialized 
from here

The change was pointed  out by Michael Qiu in patch: 
2b039d5f20a34016ecaf9b26f8f8b6c4a81bf4b6

We can only cast void to uint8 but it would involve big endian/ little endian 
macros:

static inline uint32_t __attribute__((__may_alias__))
__rte_raw_cksum(const void *buf, size_t len, uint32_t sum)
{
/* workaround gcc strict-aliasing warning */
const uint8_t *u8 = ((const uint8_t *)buf);
while (len >= (sizeof(*u8) * 8)) {
sum += *(u8+1) << 8 | *(u8);
sum += *(u8+3) << 8 | *(u8+2);
sum += *(u8+5) << 8 | *(u8+4);
sum += *(u8+7) << 8 | *(u8+6);

len -= sizeof(*u8) * 8;
u8 += 8;
}

while (len >= 2*sizeof(*u8)) {
sum += *u8 << 8 | *(u8+1);
len -= 2*sizeof(*u8);
u8 += 2;
}

/* if length is in odd bytes */
if (len == 1)
sum += *((const uint8_t *)u8);

return sum;
}

I will research about this union usage suggested by Stephen, but for this 
moment  local typedef & __attribute__((__may_alias__)) is the most clean 
solution which work under gcc(4.4.7), clang, icc.

>-Original Message-
>From: Ananyev, Konstantin
>Sent: Tuesday, November 24, 2015 6:58 PM
>To: Mrzyglod, DanielX T 
>Cc: dev at dpdk.org
>Subject: RE: [dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict 
>aliasing
>
>Hi Daniel,
>So for my own curiosity: what went wrong here?
>Did compiler avoid to generate a code for while {} loop?
>Or something else?
>BTW, it is an internal function, so instead of introducing new typedef,
>we can just change the type of buf?
>Let say to uint8_t *?
>From gcc manual page: "  A character type may alias any other type."
>Would that work?
>Konstantin
>
>> -Original Message-
>> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Daniel Mrzyglod
>> Sent: Tuesday, November 24, 2015 4:31 PM
>> To: dev at dpdk.org
>> Subject: [dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict 
>> aliasing
>>
>> This fix is for IPv6 checksum offload error on RHEL65.
>> Any optimalisation above -O0 provide error in IPv6 checksum
>> flag "-fstrict-aliasing" is default for optimalisation above -O0.
>>
>> Fixes: 2b039d5f20a3 ("net: fix build with gcc 4.4.7 and strict aliasing")
>>
>> Signed-off-by: Daniel Mrzyglod 
>> ---
>>  lib/librte_net/rte_ip.h | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/lib/librte_net/rte_ip.h b/lib/librte_net/rte_ip.h
>> index 71c519a..5b7554a 100644
>> --- a/lib/librte_net/rte_ip.h
>> +++ b/lib/librte_net/rte_ip.h
>> @@ -169,7 +169,8 @@ __rte_raw_cksum(const void *buf, size_t len, uint32_t
>sum)
>>  {
>>  /* workaround gcc strict-aliasing warning */
>>  uintptr_t ptr = (uintptr_t)buf;
>> -const uint16_t *u16 = (const uint16_t *)ptr;
>> +typedef uint16_t __attribute__((__may_alias__)) u16_p;
>> +const u16_p *u16 = (const u16_p *)ptr;
>>
>>  while (len >= (sizeof(*u16) * 4)) {
>>  sum += u16[0];
>> --
>> 2.5.0

[dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict aliasing

[Ananyev, Konstantin]

> -Original Message-
> From: Mrzyglod, DanielX T
> Sent: Wednesday, November 25, 2015 5:07 PM
> To: Ananyev, Konstantin
> Cc: dev at dpdk.org
> Subject: RE: [dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict 
> aliasing
> 
> If we change input buf:
> xapp-gcc/include/rte_ip.h:211: error: dereferencing pointer 'u16' does break 
> strict-aliasing rules
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:204: note: initialized 
> from here
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:212: error: 
> dereferencing pointer '({anonymous})' does break strict-aliasing rules
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:212: note: initialized 
> from here
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:213: error: 
> dereferencing pointer '({anonymous})' does break strict-aliasing rules
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:213: note: initialized 
> from here
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:214: error: 
> dereferencing pointer '({anonymous})' does break strict-aliasing rules
> /root/dpdk/x86_64-native-linuxapp-gcc/include/rte_ip.h:214: note: initialized 
> from here
> 
> The change was pointed  out by Michael Qiu in patch: 
> 2b039d5f20a34016ecaf9b26f8f8b6c4a81bf4b6
> 
> We can only cast void to uint8 but it would involve big endian/ little endian 
> macros:
> 
> static inline uint32_t __attribute__((__may_alias__))
> __rte_raw_cksum(const void *buf, size_t len, uint32_t sum)
> {
>   /* workaround gcc strict-aliasing warning */
>   const uint8_t *u8 = ((const uint8_t *)buf);
>   while (len >= (sizeof(*u8) * 8)) {
>   sum += *(u8+1) << 8 | *(u8);
>   sum += *(u8+3) << 8 | *(u8+2);
>   sum += *(u8+5) << 8 | *(u8+4);
>   sum += *(u8+7) << 8 | *(u8+6);
> 
>   len -= sizeof(*u8) * 8;
>   u8 += 8;
>   }
> 
>   while (len >= 2*sizeof(*u8)) {
>   sum += *u8 << 8 | *(u8+1);
>   len -= 2*sizeof(*u8);
>   u8 += 2;
>   }
> 
>   /* if length is in odd bytes */
>   if (len == 1)
>   sum += *((const uint8_t *)u8);
> 
>   return sum;
> }

So I misinterpreted gcc manual: it only allows conversion from any other type 
to char, not visa-versa.
Sorry for wrong suggestion.

> 
> I will research about this union usage suggested by Stephen, but for this 
> moment  local typedef & __attribute__((__may_alias__)) is the
> most clean solution which work under gcc(4.4.7), clang, icc.

Ok, if there is no other known way - let's stick with your original patch. 
Konstantin

> 
> >-Original Message-
> >From: Ananyev, Konstantin
> >Sent: Tuesday, November 24, 2015 6:58 PM
> >To: Mrzyglod, DanielX T 
> >Cc: dev at dpdk.org
> >Subject: RE: [dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict 
> >aliasing
> >
> >Hi Daniel,
> >So for my own curiosity: what went wrong here?
> >Did compiler avoid to generate a code for while {} loop?
> >Or something else?
> >BTW, it is an internal function, so instead of introducing new typedef,
> >we can just change the type of buf?
> >Let say to uint8_t *?
> >From gcc manual page: "  A character type may alias any other type."
> >Would that work?
> >Konstantin
> >
> >> -Original Message-
> >> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Daniel Mrzyglod
> >> Sent: Tuesday, November 24, 2015 4:31 PM
> >> To: dev at dpdk.org
> >> Subject: [dpdk-dev] [PATCH v2] net: fix build with gcc 4.4.7 and strict 
> >> aliasing
> >>
> >> This fix is for IPv6 checksum offload error on RHEL65.
> >> Any optimalisation above -O0 provide error in IPv6 checksum
> >> flag "-fstrict-aliasing" is default for optimalisation above -O0.
> >>
> >> Fixes: 2b039d5f20a3 ("net: fix build with gcc 4.4.7 and strict aliasing")
> >>
> >> Signed-off-by: Daniel Mrzyglod 
> >> ---
> >>  lib/librte_net/rte_ip.h | 3 ++-
> >>  1 file changed, 2 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/lib/librte_net/rte_ip.h b/lib/librte_net/rte_ip.h
> >> index 71c519a..5b7554a 100644
> >> --- a/lib/librte_net/rte_ip.h
> >> +++ b/lib/librte_net/rte_ip.h
> >> @@ -169,7 +169,8 @@ __rte_raw_cksum(const void *buf, size_t len, uint32_t
> >sum)
> >>  {
> >>/* workaround gcc strict-aliasing warning */
> >>uintptr_t ptr = (uintptr_t)buf;
> >> -  const uint16_t *u16 = (const uint16_t *)ptr;
> >> +  typedef uint16_t __attribute__((__may_alias__)) u16_p;
> >> +  const u16_p *u16 = (const u16_p *)ptr;
> >>
> >>while (len >= (sizeof(*u16) * 4)) {
> >>sum += u16[0];
> >> --
> >> 2.5.0

[dpdk-dev] [PATCH] virtio: fix rx ring descriptor starvation

[Xie, Huawei]

On 11/13/2015 5:33 PM, Tom Kiely wrote:
> If all rx descriptors are processed while transient
> mbuf exhaustion is present, the rx ring ends up with
> no available descriptors. Thus no packets are received
> on that ring. Since descriptor refill is performed post
> rx descriptor processing, in this case no refill is
> ever subsequently performed resulting in permanent rx
> traffic drop.
>
> Signed-off-by: Tom Kiely 
> ---
>  drivers/net/virtio/virtio_rxtx.c |6 --
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/virtio/virtio_rxtx.c 
> b/drivers/net/virtio/virtio_rxtx.c
> index 5770fa2..a95e234 100644
> --- a/drivers/net/virtio/virtio_rxtx.c
> +++ b/drivers/net/virtio/virtio_rxtx.c
> @@ -586,7 +586,8 @@ virtio_recv_pkts(void *rx_queue, struct rte_mbuf 
> **rx_pkts, uint16_t nb_pkts)
>   if (likely(num > DESC_PER_CACHELINE))
>   num = num - ((rxvq->vq_used_cons_idx + num) % 
> DESC_PER_CACHELINE);
>  
> - if (num == 0)
> + /* Refill free descriptors even if no pkts recvd */
> + if (num == 0 && virtqueue_full(rxvq))
Should the return condition be that no used buffers and we have avail
descs in avail ring, i.e,
num == 0 && rxvq->vq_free_cnt != rxvq->vq_nentries

rather than
num == 0 && rxvq->vq_free_cnt == 0
?
>   return 0;
>  
>   num = virtqueue_dequeue_burst_rx(rxvq, rcv_pkts, len, num);
> @@ -683,7 +684,8 @@ virtio_recv_mergeable_pkts(void *rx_queue,
>  
>   virtio_rmb();
>  
> - if (nb_used == 0)
> + /* Refill free descriptors even if no pkts recvd */
> + if (nb_used == 0 && virtqueue_full(rxvq))
>   return 0;
>  
>   PMD_RX_LOG(DEBUG, "used:%d\n", nb_used);

[dpdk-dev] [PATCH] examples/bond: fix bsd compile error

[Ferruh Yigit]

Error:
== bond
  CC main.o
/.../examples/bond/main.c:431:24: error: use of undeclared identifier 'AF_INET'
if (res->ip.family == AF_INET)
  ^
1 error generated.
/.../mk/internal/rte.compile-pre.mk:126: recipe for target 'main.o' failed

AF_INET defined in sys/socket.h

This header included for Linux:
. //include/rte_ip.h
.. /usr/include/netinet/in.h
... /usr/include/sys/socket.h

But not for FreeBSD:
. //include/rte_ip.h
.. /usr/include/netinet/in.h
... /usr/include/machine/endian.h
... /usr/include/netinet6/in6.h
. //include/rte_tcp.h

Signed-off-by: Ferruh Yigit 
---
 examples/bond/main.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/examples/bond/main.c b/examples/bond/main.c
index 4622283..b2d5ac2 100644
--- a/examples/bond/main.c
+++ b/examples/bond/main.c
@@ -75,6 +75,7 @@
 #include 
 #include 
 #include 
+#include 

 #include 
 #include 
-- 
2.5.0

[dpdk-dev] [PATCH] cryptodev: mark experimental state

[Thomas Monjalon]

The crypto API is in an early state.
It requires more discussions and experiments to declare it stable,
as discussed in http://dpdk.org/ml/archives/dev/2015-November/028634.html

A documentation section will be required in the guides.

Signed-off-by: Thomas Monjalon 
---
 MAINTAINERS|  2 +-
 config/common_bsdapp   |  1 +
 config/common_linuxapp |  1 +
 doc/guides/contributing/versioning.rst |  1 +
 doc/guides/rel_notes/release_2_2.rst   | 10 ++
 lib/librte_cryptodev/rte_cryptodev.h   |  3 +++
 6 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/MAINTAINERS b/MAINTAINERS
index 7f3d967..460245b 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -219,7 +219,7 @@ M: Thomas Monjalon 
 F: lib/librte_ether/
 F: scripts/test-null.sh

-Crypto API
+Crypto API - EXPERIMENTAL
 M: Declan Doherty 
 F: lib/librte_cryptodev/
 F: app/test/test_cryptodev*
diff --git a/config/common_bsdapp b/config/common_bsdapp
index 56020b6..3286481 100644
--- a/config/common_bsdapp
+++ b/config/common_bsdapp
@@ -310,6 +310,7 @@ CONFIG_RTE_PMD_PACKET_PREFETCH=y

 #
 # Compile generic crypto device library
+# EXPERIMENTAL: API may change without prior notice
 #
 CONFIG_RTE_LIBRTE_CRYPTODEV=y
 CONFIG_RTE_LIBRTE_CRYPTODEV_DEBUG=n
diff --git a/config/common_linuxapp b/config/common_linuxapp
index eaad8d6..2866986 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -319,6 +319,7 @@ CONFIG_RTE_PMD_PACKET_PREFETCH=y

 #
 # Compile generic crypto device library
+# EXPERIMENTAL: API may change without prior notice
 #
 CONFIG_RTE_LIBRTE_CRYPTODEV=y
 CONFIG_RTE_LIBRTE_CRYPTODEV_DEBUG=n
diff --git a/doc/guides/contributing/versioning.rst 
b/doc/guides/contributing/versioning.rst
index 8a739dd..653c7d0 100644
--- a/doc/guides/contributing/versioning.rst
+++ b/doc/guides/contributing/versioning.rst
@@ -13,6 +13,7 @@ General Guidelines
 --

 #. Whenever possible, ABI should be preserved
+#. The libraries marked in experimental state may change without constraint.
 #. The addition of symbols is generally not problematic
 #. The modification of symbols can generally be managed with versioning
 #. The removal of symbols generally is an ABI break and requires bumping of the
diff --git a/doc/guides/rel_notes/release_2_2.rst 
b/doc/guides/rel_notes/release_2_2.rst
index 8c77768..d0a9955 100644
--- a/doc/guides/rel_notes/release_2_2.rst
+++ b/doc/guides/rel_notes/release_2_2.rst
@@ -17,6 +17,16 @@ New Features

 * **Added keepalive support to EAL and example application.**

+* **Added experimental cryptodev API**
+
+  The cryptographic processing of packet is provided as a preview
+  with two drivers for:
+
+  * Intel QuickAssist devices
+  * Intel AES-NI multi-buffer library
+
+  Due to its experimental state, the API may change without prior notice.
+
 * **Added ethdev API to support IEEE1588.**

   Added functions to read, write and adjust system time in the NIC.
diff --git a/lib/librte_cryptodev/rte_cryptodev.h 
b/lib/librte_cryptodev/rte_cryptodev.h
index 04bade7..aa9f785 100644
--- a/lib/librte_cryptodev/rte_cryptodev.h
+++ b/lib/librte_cryptodev/rte_cryptodev.h
@@ -39,6 +39,9 @@
  *
  * Defines RTE Crypto Device APIs for the provisioning of cipher and
  * authentication operations.
+ *
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
  */

 #ifdef __cplusplus
-- 
2.5.2

[dpdk-dev] [PATCH v2] examples/bond: fix bsd compile error

[Ferruh Yigit]

Error:
== bond
  CC main.o
/.../examples/bond/main.c:431:24: error: use of undeclared identifier 'AF_INET'
if (res->ip.family == AF_INET)
  ^
1 error generated.
/.../mk/internal/rte.compile-pre.mk:126: recipe for target 'main.o' failed

AF_INET defined in sys/socket.h

This header included for Linux:
. //include/rte_ip.h
.. /usr/include/netinet/in.h
... /usr/include/sys/socket.h

But not for FreeBSD:
. //include/rte_ip.h
.. /usr/include/netinet/in.h
... /usr/include/machine/endian.h
... /usr/include/netinet6/in6.h
. //include/rte_tcp.h

Signed-off-by: Ferruh Yigit 
---
 examples/bond/main.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/examples/bond/main.c b/examples/bond/main.c
index 4622283..19f4f05 100644
--- a/examples/bond/main.c
+++ b/examples/bond/main.c
@@ -45,6 +45,7 @@
 #include 
 #include 
 #include 
+#include 

 #include 
 #include 
-- 
2.5.0

[dpdk-dev] [PATCH v8 00/10] Crypto API and device framework

[Thomas Monjalon]

2015-11-25 13:25, Declan Doherty:
> This series of patches defines a set of application burst oriented APIs for
> asynchronous symmetric cryptographic functions within DPDK. It also contains a
> poll mode driver cryptographic device framework for the implementation of
> crypto devices within DPDK.
> 
> In the patch set we also have included 2 reference implementations of crypto
> PMDs. Currently both implementations support AES-CBC with
> HMAC_SHA1/SHA256/SHA512 authentication operations. The first device is a 
> purely
> software PMD based on Intel's multi-buffer library, which utilises both
> AES-NI instructions and vector operations to accelerate crypto operations and
> the second PMD utilises Intel's Quick Assist Technology (on DH895xxC) to
> provide hardware accelerated crypto operations.

After rebase, small fixes in configs and MAINTAINERS,
Applied, thanks

It is marked as experimental in the documentation and configs.
The API needs to be more tested, discussed and documented.
So it is not stable and no deprecation process is needed to make some changes.

[dpdk-dev] rte_eth_tx_queue_setup() failing (error -22) when setting up tx queues on a VMXNET3 port...

[Montorsi, Francesco]

Hi Stephen,

> -Original Message-
> From: Stephen Hemminger [mailto:stephen at networkplumber.org]
> 
> If you read the source, you will see that it there are log messages enabled if
> you configure with LIBRTE_ETHDEV_DEBUG enabled and rebuild DPDK. And
> there are log messages for VMXNETE3 controlled by
> LIBRTE_VMXNET3_DEBUG_INIT

Thanks for this hint. I didn't know actually where to read... anyway enabling 
these additional debug messages I found the cause of the problem: the minimal 
TX queue ring size for VMXNET3 is 512; I was passing 64.
Fixed this problem, I'm now on the next one I discovered that apparently 
the driver for VMXNET3 supports only a limited set of functions:

(gdb) p *dev->dev_ops
$3 = {dev_configure = 0x75cda5 , dev_start = 0x75d653 
, dev_stop = 0x75d799 , dev_set_link_up = 
0, dev_set_link_down = 0,
  dev_close = 0x75d8ab , promiscuous_enable = 0x75dd12 
, promiscuous_disable = 0x75ddf6 
,
  allmulticast_enable = 0x75df0c , 
allmulticast_disable = 0x75df40 , link_update 
= 0x75dbd9 ,
  stats_get = 0x75d8fe , stats_reset = 0, xstats_get = 
0, xstats_reset = 0, queue_stats_mapping_set = 0, dev_infos_get = 0x75db7f 
,
  mtu_set = 0, vlan_filter_set = 0x75df74 , 
vlan_tpid_set = 0, vlan_strip_queue_set = 0, vlan_offload_set = 0x75e455 
,
  vlan_pvid_set = 0, rx_queue_start = 0, rx_queue_stop = 0, tx_queue_start = 0, 
tx_queue_stop = 0, rx_queue_setup = 0x75bfb8 ,
  rx_queue_release = 0x74b069 , rx_queue_count = 
0, rx_descriptor_done = 0, rx_queue_intr_enable = 0, rx_queue_intr_disable = 0,
  tx_queue_setup = 0x75bbc6 , tx_queue_release = 
0x74b03c , dev_led_on = 0, dev_led_off = 0, 
flow_ctrl_get = 0,
  flow_ctrl_set = 0, priority_flow_ctrl_set = 0, mac_addr_remove = 0, 
mac_addr_add = 0, mac_addr_set = 0, uc_hash_table_set = 0, 
uc_all_hash_table_set = 0, mirror_rule_set = 0,
  mirror_rule_reset = 0, set_vf_rx_mode = 0, set_vf_rx = 0, set_vf_tx = 0, 
set_vf_vlan_filter = 0, udp_tunnel_add = 0, udp_tunnel_del = 0, 
set_queue_rate_limit = 0, set_vf_rate_limit = 0,
  fdir_add_signature_filter = 0, fdir_update_signature_filter = 0, 
fdir_remove_signature_filter = 0, fdir_infos_get = 0, fdir_add_perfect_filter = 
0, fdir_update_perfect_filter = 0,
  fdir_remove_perfect_filter = 0, fdir_set_masks = 0, reta_update = 0, 
reta_query = 0, get_reg_length = 0, get_reg = 0, get_eeprom_length = 0, 
get_eeprom = 0, set_eeprom = 0,
  rss_hash_update = 0, rss_hash_conf_get = 0, filter_ctrl = 0, set_mc_addr_list 
= 0, timesync_enable = 0, timesync_disable = 0, timesync_read_rx_timestamp = 0,
  timesync_read_tx_timestamp = 0}


Since I was using the rte_eth_rx_queue_count() function in one place, and 
VMXNET3 does not support it, I'm getting a SEGFAULT.
So next question is:  is user's task to check for validity of pointers inside 
dev_ops before calling driver functions? Because rte_eth_rx_queue_count() and 
companion funcitons have no safety checks apparently (!!!)

Thanks!

Francesco

[dpdk-dev] Issues with running multicore

[Sugumaran, Varthamanan]

Hi All,
Iam trying to create multi core application by mapping separate function for 
each core.
I had followed Multi process example and I had used pthread.
The use case is:

1.  In a single process, launch pthread and run rte_eal_init

2.  Once  rte_eal_init is done, create 3 pthreads

3.  Launch master core.

4.  In each of 3 pthread, launch slave core.

The issue is, I don't see 4 cores running. Instead I see only single core 
running.

Here is the Pseudo code:

Init_pmd{
   ret = rte_eal_init(args, argv) ==> I had set the number of cores to 4 in the 
params.

pthread_create(&thread1, NULL, (void*)threadFun1, NULL);
pthread_create(&thread2, NULL, (void*)threadFun2, NULL);
pthread_create(&thread3, NULL, (void*)threadFun3, NULL);

   rte_eal_mp_remote_launch(launch_pmd_single_lcore, NULL, CALL_MASTER);

}

Iam mapping each core to threadFun1, threadFun2 and threadFun3.

threadFun1
{
   RTE_LCORE_FOREACH_SLAVE(lcore_id)
   {
  if( lcore_id == 1 )
 rte_eal_remote_launch(threadFun1_launch_one_lcore, NULL, lcore_id);
   }
}

Has anyone seen this issue?
Please let me know if iam missing something?

Thanks
Vartha

[dpdk-dev] rte_eth_tx_queue_setup() failing (error -22) when setting up tx queues on a VMXNET3 port...

[Thomas Monjalon]

2015-11-25 17:55, Montorsi, Francesco:
> Since I was using the rte_eth_rx_queue_count() function in one place, and 
> VMXNET3 does not support it, I'm getting a SEGFAULT.
> So next question is:  is user's task to check for validity of pointers inside 
> dev_ops before calling driver functions? Because rte_eth_rx_queue_count() and 
> companion funcitons have no safety checks apparently (!!!)

Bruce already sent a fix for this issue.
It should be applied on HEAD in few minutes.

[dpdk-dev] [PATCH] net: fix build with gcc 4.4.7 and strict aliasing

[Ananyev, Konstantin]

> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Daniel Mrzyglod
> Sent: Tuesday, November 24, 2015 3:13 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH] net: fix build with gcc 4.4.7 and strict aliasing
> 
> This is fix for GCC 4.4.7.
> flag "-fstrict-aliasing" is default for optimalisation above -O0.
> 
> Fixes: 2b039d5f20a3 ("net: fix build with gcc 4.4.7 and strict aliasing")
> 
> Signed-off-by: Daniel Mrzyglod 
> ---

Acked-by: Konstantin Ananyev 

> 2.5.0

[dpdk-dev] [PATCH] examples/bond: add header to support freebsd compilation

[Daniel Mrzyglod]

definition of 'AF_INET' enum was missing - is available in 

Signed-off-by: Daniel Mrzyglod 
---
 examples/bond/main.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/examples/bond/main.c b/examples/bond/main.c
index 4622283..53bd044 100644
--- a/examples/bond/main.c
+++ b/examples/bond/main.c
@@ -33,6 +33,7 @@

 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
-- 
2.5.0

[dpdk-dev] [PATCH] examples/bond: add header to support freebsd compilation

[Thomas Monjalon]

2015-11-25 19:03, Daniel Mrzyglod:
> definition of 'AF_INET' enum was missing - is available in 
> 
> Signed-off-by: Daniel Mrzyglod 

It is definitely the right fix as Ferruh submitted the same one
less than one hour ago.
Should we understand it is an ack?

It seems Ferruh was hesitating about the line where inserting the
include. I would say I prefer your choice :)

[dpdk-dev] [PATCH v5 1/2] ethdev: remove duplicated debug functions

[Thomas Monjalon]

2015-11-24 17:37, Bruce Richardson:
> --- a/lib/librte_ether/Makefile
> +++ b/lib/librte_ether/Makefile
> @@ -41,7 +41,7 @@ CFLAGS += $(WERROR_FLAGS)
>  
>  EXPORT_MAP := rte_ether_version.map
>  
> -LIBABIVER := 2
> +LIBABIVER := 3

This number has already been increased at the beginning of this release cycle.
I will drop this change.

[dpdk-dev] [PATCH v5 0/2] ethdev: debug code cleanup

[Thomas Monjalon]

2015-11-24 17:37, Bruce Richardson:
> This patchset performs two cleanups:
> 1. Four functions in ethdev.c which were enabled for debug only have been
>   merged into their inlined header-file counterparts. This change required 
> that
>   a number of macros be renamed and moved to the header file too. The macro 
> changes
>   are in patches 1 & 2, and the elimination of the separate debug fns are in 
> patch 3.
> 2. Checks for valid function pointers are added to the API calls for reading
>   the descriptor ring count, and checking for a valid descriptor. This is 
> because
>   these functions are not implemented by most drivers, and so it's far safer 
> to
>   have the check.
> 
> NOTE: This patchset now depends upon the cryptodev patchset

Applied with small changes, thanks

[dpdk-dev] [RFC PATCH 2/2] lib/librte_eal: Remove unnecessary hugepage zero-filling

[Xie, Huawei]

On 11/23/2015 2:52 PM, Stephen Hemminger wrote:
> On Mon, 23 Nov 2015 05:05:21 +
> "Xie, Huawei"  wrote:
>
>> On 11/23/2015 12:07 PM, Stephen Hemminger wrote:
>>> On Mon, 23 Nov 2015 03:46:31 +
>>> "Xie, Huawei"  wrote:
>>>
> Why cannot we rely on the kernel zeroing the memory ?
> If that behavior were to change, then we can zero out the memory
> ourselves.  
 It is undocumented kernel behavior. My opinion is if not a big burden,
 zero out the needed memory ourselves, otherwise resort to this kernel
 behavior.
>>> Really, I think it is more an oversight of missing documentation,
>>> the kernel has always (and will continue) to zero out memory that is given
>>> to a process. If it didn't it would be a massive security hole.
>> Agree. I believe this behavior will not change in future. For the
>> security issue, kernel could also set all bits like to 1. Just wonder if
>> this is best practice and whether there are other user space programs
>> rely on this behavior.
>>
> Glibc almost certainly depends on this, because heap is grown by mmaping 
> addtional
> memory.
Thanks. It is OK, but still don't feel good, :). It is like that we
require the clear_user_page(and other unix, windows equivalent) is
always memset(ptr, 0 , PAGE_SIZE). To me, memset(ptr, 1, PAGE_SIZE)
doesn't make difference.

[dpdk-dev] [PATCH v8 1/9] nfp: basic initialization

[Thomas Monjalon]

2015-11-25 16:19, Alejandro.Lucero:
>  config/common_linuxapp   |6 +
>  doc/guides/rel_notes/release_2_2.rst |4 +
>  drivers/net/Makefile |1 +
>  drivers/net/nfp/Makefile |   54 +++
>  drivers/net/nfp/nfp_net.c|  699 
> ++
>  drivers/net/nfp/nfp_net_ctrl.h   |  324 
>  drivers/net/nfp/nfp_net_logs.h   |   75 
>  drivers/net/nfp/nfp_net_pmd.h|  453 ++
>  mk/rte.app.mk|1 +

It does not compile as a shared library:
Must Specify a librte_pmd_nfp.so..1 ABI version

[dpdk-dev] [PATCH v8 1/9] nfp: basic initialization

[Thomas Monjalon]

2015-11-25 16:19, Alejandro.Lucero:
> From: "Alejandro.Lucero" 
> 
> Signed-off-by: Alejandro.Lucero 
> Signed-off-by: Rolf.Neugebauer 

Please could you remove the dot in the names?