Re: [dpdk-dev] [PATCH v3 0/8] Add crypto PMD optimized for ARMv8

[Hemant Agrawal]

On 1/4/2017 11:03 PM, zbigniew.bo...@caviumnetworks.com wrote:

From: Zbigniew Bodek 

Introduce crypto poll mode driver using ARMv8
cryptographic extensions. This PMD is optimized
to provide performance boost for chained
crypto operations processing, such as:
* encryption + HMAC generation
* decryption + HMAC validation.
In particular, cipher only or hash only
operations are not provided.


Do you have a plan to add the crypto only, auth/hash only support into 
this driver?
Also, do you plan to add additional cases w.r.t supported by other 
crypto driver?



Performance gain can be observed in tests
against OpenSSL PMD which also uses ARM
crypto extensions for packets processing.

Exemplary crypto performance tests comparison:

cipher_hash. cipher algo: AES_CBC
auth algo: SHA1_HMAC cipher key size=16.
burst_size: 64 ops

ARMv8 PMD improvement over OpenSSL PMD
(Optimized for ARMv8 cipher only and hash
only cases):

Buffer
Size(B)   OPS(M)  Throughput(Gbps)
64729 %742 %
128   577 %592 %
256   483 %476 %
512   336 %351 %
768   300 %286 %
1024  263 %250 %
1280  225 %229 %
1536  214 %213 %
1792  186 %203 %
2048  200 %193 %

The driver currently supports AES-128-CBC
in combination with: SHA256 HMAC and SHA1 HMAC.
The core crypto functionality of this driver is
provided by the external armv8_crypto library
that can be downloaded from the Cavium repository:
https://github.com/caviumnetworks/armv8_crypto

CPU compatibility with this virtual device
is detected in run-time and virtual crypto
device will not be created if CPU doesn't
provide AES, SHA1, SHA2 and NEON.

The functionality and performance of this
code can be tested using generic test application
with the following commands:
* cryptodev_sw_armv8_autotest
* cryptodev_sw_armv8_perftest
New test vectors and cases have been added
to the general pool. In particular SHA1 and
SHA256 HMAC for short cases were introduced.
This is because low-level ARM assembly code
is using different code paths for long and
short data sets, so in order to test the
mentioned driver correctly, two different
data sets need to be provided.

---
v3:
* Addressed review remarks
* Moved low-level assembly code to the external library
* Removed SHA256 MAC cases
* Various fixes: interface to the library, digest destination
  and source address interpreting, missing mbuf manipulations.

v2:
* Fixed checkpatch warnings
* Divide patches into smaller logical parts

Zbigniew Bodek (8):
  mk: fix build of assembly files for ARM64
  lib: add cryptodev type for the upcoming ARMv8 PMD
  crypto/armv8: add PMD optimized for ARMv8 processors
  mk/crypto/armv8: add PMD to the build system
  doc/armv8: update documentation about crypto PMD
  crypto/armv8: enable ARMv8 PMD in the configuration
  crypto/armv8: update MAINTAINERS entry for ARMv8 crypto
  app/test: add ARMv8 crypto tests and test vectors

 MAINTAINERS|   6 +
 app/test/test_cryptodev.c  |  63 ++
 app/test/test_cryptodev_aes_test_vectors.h | 144 +++-
 app/test/test_cryptodev_blockcipher.c  |   4 +
 app/test/test_cryptodev_blockcipher.h  |   1 +
 app/test/test_cryptodev_perf.c | 480 +
 config/common_base |   6 +
 doc/guides/cryptodevs/armv8.rst|  96 +++
 doc/guides/cryptodevs/index.rst|   1 +
 doc/guides/rel_notes/release_17_02.rst |   5 +
 drivers/crypto/Makefile|   1 +
 drivers/crypto/armv8/Makefile  |  73 ++
 drivers/crypto/armv8/rte_armv8_pmd.c   | 926 +
 drivers/crypto/armv8/rte_armv8_pmd_ops.c   | 369 ++
 drivers/crypto/armv8/rte_armv8_pmd_private.h   | 211 ++
 drivers/crypto/armv8/rte_armv8_pmd_version.map |   3 +
 lib/librte_cryptodev/rte_cryptodev.h   |   3 +
 mk/arch/arm64/rte.vars.mk  |   1 -
 mk/rte.app.mk  |   2 +
 mk/toolchain/gcc/rte.vars.mk   |   6 +-
 20 files changed, 2390 insertions(+), 11 deletions(-)
 create mode 100644 doc/guides/cryptodevs/armv8.rst
 create mode 100644 drivers/crypto/armv8/Makefile
 create mode 100644 drivers/crypto/armv8/rte_armv8_pmd.c
 create mode 100644 drivers/crypto/armv8/rte_armv8_pmd_ops.c
 create mode 100644 drivers/crypto/armv8/rte_armv8_pmd_private.h
 create mode 100644 drivers/crypto/armv8/rte_armv8_pmd_version.map

Re: [dpdk-dev] [PATCH v3 1/8] mk: fix build of assembly files for ARM64

[Hemant Agrawal]

On 1/4/2017 11:03 PM, zbigniew.bo...@caviumnetworks.com wrote:

From: Zbigniew Bodek 

Avoid using incorrect assembler (nasm) and unsupported flags
when building for ARM64.

Fixes:  af75078fece3 ("first public release")
b3ce00e5fe36 ("mk: introduce ARMv8 architecture")

Signed-off-by: Zbigniew Bodek 
---
 mk/arch/arm64/rte.vars.mk| 1 -
 mk/toolchain/gcc/rte.vars.mk | 6 --
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/mk/arch/arm64/rte.vars.mk b/mk/arch/arm64/rte.vars.mk
index c168426..3b1178a 100644
--- a/mk/arch/arm64/rte.vars.mk
+++ b/mk/arch/arm64/rte.vars.mk
@@ -53,7 +53,6 @@ CROSS ?=

 CPU_CFLAGS  ?=
 CPU_LDFLAGS ?=
-CPU_ASFLAGS ?= -felf

 export ARCH CROSS CPU_CFLAGS CPU_LDFLAGS CPU_ASFLAGS

diff --git a/mk/toolchain/gcc/rte.vars.mk b/mk/toolchain/gcc/rte.vars.mk
index ff70f3d..94f6412 100644
--- a/mk/toolchain/gcc/rte.vars.mk
+++ b/mk/toolchain/gcc/rte.vars.mk
@@ -41,9 +41,11 @@
 CC= $(CROSS)gcc
 KERNELCC  = $(CROSS)gcc
 CPP   = $(CROSS)cpp
-# for now, we don't use as but nasm.
-# AS  = $(CROSS)as
+ifeq ($(CONFIG_RTE_ARCH_X86),y)
 AS= nasm
+else
+AS= $(CROSS)as
+endif
 AR= $(CROSS)ar
 LD= $(CROSS)ld
 OBJCOPY   = $(CROSS)objcopy



you may add:
Acked-by: Hemant Agrawal 

on a side note=> This patch is not related to this patch series anymore.

Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device memory read/write API

[Jerin Jacob]

On Thu, Jan 12, 2017 at 07:09:22PM +, Ferruh Yigit wrote:
> Hi Jerin,
> 
> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> <...>
> 
> > +#include 
> > +
> >  /* CSR write macro */
> > -#define ADF_CSR_WR(csrAddr, csrOffset, val) \
> > -   (void)((*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)) \
> > -   = (val)))
> > +#define ADF_CSR_WR(csrAddr, csrOffset, val)\
> > +   rte_write32(val, (((uint8_t *)csrAddr) + csrOffset))
> 
> For IA, this update introduces an extra compiler barrier (rte_io_wmb()),
> which is indeed not a must, is this correct?

AFAIK, Compiler barrier is required for IA. I am not an IA expert, if
someone thinks it needs to changed then I can fix it in following commit
in this patch series by making rte_io_wmb() and rte_io_rmb() as empty.

Let me know.

AFAIK, Linux kernel code has a barrier in readl/writel for IA.

Typically we don't use any non relaxed versions in fast path.In fast
typically all the drivers has explicit write barrier for doorbell write
and followed by a relaxed version of write. IMO, In any event, it won't
generate performance regression.

[dpdk-master] $ git show
70c343bdc8c33a51a9db23cd58122bdfc120a58f
commit 70c343bdc8c33a51a9db23cd58122bdfc120a58f
Author: Jerin Jacob 
Date:   Mon Dec 5 06:36:49 2016 +0530

eal/x86: define I/O device memory barriers for IA

The patch does not provide any functional change for IA.
I/O barriers are mapped to existing smp barriers.

CC: Bruce Richardson 
CC: Konstantin Ananyev 
Signed-off-by: Jerin Jacob 

diff --git a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
index 00b1cdf..4eac666 100644
--- a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
+++ b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
@@ -61,6 +61,12 @@ extern "C" {
 
 #define rte_smp_rmb() rte_compiler_barrier()
 
+#define rte_io_mb() rte_mb()
+
+#define rte_io_wmb() rte_compiler_barrier()
+
+#define rte_io_rmb() rte_compiler_barrier()
+
 /*- 16 bit atomic operations
 * -*/
 
 #ifndef RTE_FORCE_INTRINSICS

> 
> If so, does it make sense to override these functions for x86, and make
> rte_writeX = rte_writeX_relaxed
> rte_readX = rte_readX_relaxed
> 
> >  
> >  /* CSR read macro */
> > -#define ADF_CSR_RD(csrAddr, csrOffset) \
> > -   (*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)))
> > +#define ADF_CSR_RD(csrAddr, csrOffset) \
> > +   rte_read32uint8_t *)csrAddr) + csrOffset))
> 
> This patchset both introduces new rte_readX/rte_writeX functions, also
> applies them into drivers.
> 
> While applying them, it changes the behavior.
> Like above code was doing a read, but after update it does read and
> read_memory_barrier.
> 
> What do you think this patchset updates usage in a manner that keeps
> behavior exact same. Like using rte_read32_relaxed for this case.
> And doing architecture related updates in a different patchset?

Need to use rte_read32 at this commit otherwise it will break for ARM.
That's was all point for this patchset.
For performance regression, we can always verify by taking delta
between this changeset and the previous changeset. If you think, I need
to make rte_io_wmb()/rte_io_rmb() as empty for IA then I could do that
as well.


> 
> This both makes easy to see architecture specific updates, and makes
> easy to trace any possible performance issues by this patchset.
> 
> >  
> >  #define ADF_BANK_INT_SRC_SEL_MASK_0 0x444CUL
> >  #define ADF_BANK_INT_SRC_SEL_MASK_X 0xUL
> > 
> 

[dpdk-dev] [PATCH v6 00/18] net/ixgbe: Consistent filter API

[Wei Zhao]

The patches mainly finish following functions:
1) Store and restore all kinds of filters.
2) Parse all kinds of filters.
3) Add flow validate function.
4) Add flow create function.
5) Add flow destroy function.
6) Add flow flush function.

v2 changes:
 fix git log error
 Modify some function call relationship
 Change return value type of all parse flow functions
 Update error info for all flow ops
 Add ixgbe_filterlist_flush to flush flows and rules created

v3 change:
 add new file ixgbe_flow.c to store generic API parser related functions
 add more comment about pattern and action rules
 add attr check in parser functions
 change struct name ixgbe_flow to rte_flow
 change SYN to TCP SYN
 change to use memset initizlize struct ixgbe_filter_info
 break down filter uninit process to 3 indepedent functions in 
eth_ixgbe_dev_uninit()
 change struct rte_flow_item_nvgre definition
 change struct rte_flow_item_e_tag definition
 fix one bug in function ixgbe_dev_filter_ctrl
 add goto in function ixgbe_flow_create
 delete some useless initialization 
 eliminate some git log check warning

v4 change:
 fix some check patch warning

v5 change:
 fix some git log warning

v6 change:
 add more comments on pattern rule example, supply the usage of spec
 last and mask

zhao wei (18):
  net/ixgbe: store TCP SYN filter
  net/ixgbe: store flow director filter
  net/ixgbe: store L2 tunnel filter
  net/ixgbe: restore n-tuple filter
  net/ixgbe: restore ether type filter
  net/ixgbe: restore TCP SYN filter
  net/ixgbe: restore flow director filter
  net/ixgbe: restore L2 tunnel filter
  net/ixgbe: store and restore L2 tunnel configuration
  net/ixgbe: flush all the filters
  net/ixgbe: parse n-tuple filter
  net/ixgbe: parse ethertype filter
  net/ixgbe: parse TCP SYN filter
  net/ixgbe: parse L2 tunnel filter
  net/ixgbe: parse flow director filter
  net/ixgbe: create consistent filter
  net/ixgbe: destroy consistent filter
  net/ixgbe: flush all the filter list

 drivers/net/ixgbe/Makefile   |2 +
 drivers/net/ixgbe/ixgbe_ethdev.c |  667 +++--
 drivers/net/ixgbe/ixgbe_ethdev.h |  203 ++-
 drivers/net/ixgbe/ixgbe_fdir.c   |  407 --
 drivers/net/ixgbe/ixgbe_flow.c   | 2878 ++
 drivers/net/ixgbe/ixgbe_pf.c |   26 +-
 lib/librte_ether/rte_flow.h  |   48 +
 7 files changed, 4020 insertions(+), 211 deletions(-)
 create mode 100644 drivers/net/ixgbe/ixgbe_flow.c

Acked-by: Beilei Xing 
Acked-by: Wei Dai 
-- 
2.5.5

[dpdk-dev] [PATCH v6 01/18] net/ixgbe: store TCP SYN filter

[Wei Zhao]

Add support for storing TCP SYN filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 18 +-
 drivers/net/ixgbe/ixgbe_ethdev.h |  2 ++
 2 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index b7ddd4f..719eddd 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -1272,10 +1272,12 @@ eth_ixgbe_dev_init(struct rte_eth_dev *eth_dev)
/* enable support intr */
ixgbe_enable_intr(eth_dev);
 
+   /* initialize filter info */
+   memset(filter_info, 0,
+  sizeof(struct ixgbe_filter_info));
+
/* initialize 5tuple filter list */
TAILQ_INIT(&filter_info->fivetuple_list);
-   memset(filter_info->fivetuple_mask, 0,
-  sizeof(uint32_t) * IXGBE_5TUPLE_ARRAY_SIZE);
 
return 0;
 }
@@ -5603,15 +5605,18 @@ ixgbe_syn_filter_set(struct rte_eth_dev *dev,
bool add)
 {
struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
+   uint32_t syn_info;
uint32_t synqf;
 
if (filter->queue >= IXGBE_MAX_RX_QUEUE_NUM)
return -EINVAL;
 
-   synqf = IXGBE_READ_REG(hw, IXGBE_SYNQF);
+   syn_info = filter_info->syn_info;
 
if (add) {
-   if (synqf & IXGBE_SYN_FILTER_ENABLE)
+   if (syn_info & IXGBE_SYN_FILTER_ENABLE)
return -EINVAL;
synqf = (uint32_t)(((filter->queue << 
IXGBE_SYN_FILTER_QUEUE_SHIFT) &
IXGBE_SYN_FILTER_QUEUE) | IXGBE_SYN_FILTER_ENABLE);
@@ -5621,10 +5626,13 @@ ixgbe_syn_filter_set(struct rte_eth_dev *dev,
else
synqf &= ~IXGBE_SYN_FILTER_SYNQFP;
} else {
-   if (!(synqf & IXGBE_SYN_FILTER_ENABLE))
+   synqf = IXGBE_READ_REG(hw, IXGBE_SYNQF);
+   if (!(syn_info & IXGBE_SYN_FILTER_ENABLE))
return -ENOENT;
synqf &= ~(IXGBE_SYN_FILTER_QUEUE | IXGBE_SYN_FILTER_ENABLE);
}
+
+   filter_info->syn_info = synqf;
IXGBE_WRITE_REG(hw, IXGBE_SYNQF, synqf);
IXGBE_WRITE_FLUSH(hw);
return 0;
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index 69b276f..90a89ec 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -262,6 +262,8 @@ struct ixgbe_filter_info {
/* Bit mask for every used 5tuple filter */
uint32_t fivetuple_mask[IXGBE_5TUPLE_ARRAY_SIZE];
struct ixgbe_5tuple_filter_list fivetuple_list;
+   /* store the SYN filter info */
+   uint32_t syn_info;
 };
 
 /*
-- 
2.5.5

[dpdk-dev] [PATCH v6 02/18] net/ixgbe: store flow director filter

[Wei Zhao]

Add support for storing flow director filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |  64 
 drivers/net/ixgbe/ixgbe_ethdev.h |  19 ++-
 drivers/net/ixgbe/ixgbe_fdir.c   | 105 ++-
 3 files changed, 185 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 719eddd..9796c4f 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -60,6 +60,7 @@
 #include 
 #include 
 #include 
+#include 
 
 #include "ixgbe_logs.h"
 #include "base/ixgbe_api.h"
@@ -165,6 +166,8 @@ enum ixgbevf_xcast_modes {
 
 static int eth_ixgbe_dev_init(struct rte_eth_dev *eth_dev);
 static int eth_ixgbe_dev_uninit(struct rte_eth_dev *eth_dev);
+static int ixgbe_fdir_filter_init(struct rte_eth_dev *eth_dev);
+static int ixgbe_fdir_filter_uninit(struct rte_eth_dev *eth_dev);
 static int  ixgbe_dev_configure(struct rte_eth_dev *dev);
 static int  ixgbe_dev_start(struct rte_eth_dev *dev);
 static void ixgbe_dev_stop(struct rte_eth_dev *dev);
@@ -1279,6 +1282,9 @@ eth_ixgbe_dev_init(struct rte_eth_dev *eth_dev)
/* initialize 5tuple filter list */
TAILQ_INIT(&filter_info->fivetuple_list);
 
+   /* initialize flow director filter list & hash */
+   ixgbe_fdir_filter_init(eth_dev);
+
return 0;
 }
 
@@ -1320,9 +1326,67 @@ eth_ixgbe_dev_uninit(struct rte_eth_dev *eth_dev)
rte_free(eth_dev->data->hash_mac_addrs);
eth_dev->data->hash_mac_addrs = NULL;
 
+   /* remove all the fdir filters & hash */
+   ixgbe_fdir_filter_uninit(eth_dev);
+
return 0;
 }
 
+static int ixgbe_fdir_filter_uninit(struct rte_eth_dev *eth_dev)
+{
+   struct ixgbe_hw_fdir_info *fdir_info =
+   IXGBE_DEV_PRIVATE_TO_FDIR_INFO(eth_dev->data->dev_private);
+   struct ixgbe_fdir_filter *fdir_filter;
+
+   if (fdir_info->hash_map)
+   rte_free(fdir_info->hash_map);
+   if (fdir_info->hash_handle)
+   rte_hash_free(fdir_info->hash_handle);
+
+   while ((fdir_filter = TAILQ_FIRST(&fdir_info->fdir_list))) {
+   TAILQ_REMOVE(&fdir_info->fdir_list,
+fdir_filter,
+entries);
+   rte_free(fdir_filter);
+   }
+
+   return 0;
+}
+
+static int ixgbe_fdir_filter_init(struct rte_eth_dev *eth_dev)
+{
+   struct ixgbe_hw_fdir_info *fdir_info =
+   IXGBE_DEV_PRIVATE_TO_FDIR_INFO(eth_dev->data->dev_private);
+   char fdir_hash_name[RTE_HASH_NAMESIZE];
+   struct rte_hash_parameters fdir_hash_params = {
+   .name = fdir_hash_name,
+   .entries = IXGBE_MAX_FDIR_FILTER_NUM,
+   .key_len = sizeof(union ixgbe_atr_input),
+   .hash_func = rte_hash_crc,
+   .hash_func_init_val = 0,
+   .socket_id = rte_socket_id(),
+   };
+
+   TAILQ_INIT(&fdir_info->fdir_list);
+   snprintf(fdir_hash_name, RTE_HASH_NAMESIZE,
+"fdir_%s", eth_dev->data->name);
+   fdir_info->hash_handle = rte_hash_create(&fdir_hash_params);
+   if (!fdir_info->hash_handle) {
+   PMD_INIT_LOG(ERR, "Failed to create fdir hash table!");
+   return -EINVAL;
+   }
+   fdir_info->hash_map = rte_zmalloc("ixgbe",
+ sizeof(struct ixgbe_fdir_filter *) *
+ IXGBE_MAX_FDIR_FILTER_NUM,
+ 0);
+   if (!fdir_info->hash_map) {
+   PMD_INIT_LOG(ERR,
+"Failed to allocate memory for fdir hash map!");
+   return -ENOMEM;
+   }
+
+   return 0;
+}
 /*
  * Negotiate mailbox API version with the PF.
  * After reset API version is always set to the basic one (ixgbe_mbox_api_10).
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index 90a89ec..300542e 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -38,6 +38,7 @@
 #include "base/ixgbe_dcb_82598.h"
 #include "ixgbe_bypass.h"
 #include 
+#include 
 
 /* need update link, bit flag */
 #define IXGBE_FLAG_NEED_LINK_UPDATE (uint32_t)(1 << 0)
@@ -130,10 +131,11 @@
 #define IXGBE_MISC_VEC_ID   RTE_INTR_VEC_ZERO_OFFSET
 #define IXGBE_RX_VEC_START  RTE_INTR_VEC_RXTX_OFFSET
 
+#define IXGBE_MAX_FDIR_FILTER_NUM   (1024 * 32)
+
 /*
  * Information about the fdir mode.
  */
-
 struct ixgbe_hw_fdir_mask {
uint16_t vlan_tci_mask;
uint32_t src_ipv4_mask;
@@ -148,6 +150,17 @@ struct ixgbe_hw_fdir_mask {
uint8_t  tunnel_type_mask;
 };
 
+struct ixgbe_fdir_filter {
+   TAILQ_ENTRY(ixgbe_fdir_filter) entries;
+   union ixgbe_atr_input ixgbe_fdir; /* key of fdir filter*/
+   uint32_t fdirflags; /* drop or forward */
+   uint32_t fdirhash; 

[dpdk-dev] [PATCH v6 06/18] net/ixgbe: restore TCP SYN filter

[Wei Zhao]

Add support for restoring TCP SYN filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 18 ++
 1 file changed, 18 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 6c46354..6cd5975 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -7945,11 +7945,29 @@ ixgbe_ethertype_filter_restore(struct rte_eth_dev *dev)
}
 }
 
+/* restore SYN filter */
+static inline void
+ixgbe_syn_filter_restore(struct rte_eth_dev *dev)
+{
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
+   uint32_t synqf;
+
+   synqf = filter_info->syn_info;
+
+   if (synqf & IXGBE_SYN_FILTER_ENABLE) {
+   IXGBE_WRITE_REG(hw, IXGBE_SYNQF, synqf);
+   IXGBE_WRITE_FLUSH(hw);
+   }
+}
+
 static int
 ixgbe_filter_restore(struct rte_eth_dev *dev)
 {
ixgbe_ntuple_filter_restore(dev);
ixgbe_ethertype_filter_restore(dev);
+   ixgbe_syn_filter_restore(dev);
 
return 0;
 }
-- 
2.5.5

[dpdk-dev] [PATCH v6 04/18] net/ixgbe: restore n-tuple filter

[Wei Zhao]

Add support for restoring n-tuple filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 140 +--
 1 file changed, 92 insertions(+), 48 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index e63b635..1630e65 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -170,6 +170,7 @@ static int ixgbe_fdir_filter_init(struct rte_eth_dev 
*eth_dev);
 static int ixgbe_fdir_filter_uninit(struct rte_eth_dev *eth_dev);
 static int ixgbe_l2_tn_filter_init(struct rte_eth_dev *eth_dev);
 static int ixgbe_l2_tn_filter_uninit(struct rte_eth_dev *eth_dev);
+static int ixgbe_ntuple_filter_uninit(struct rte_eth_dev *eth_dev);
 static int  ixgbe_dev_configure(struct rte_eth_dev *dev);
 static int  ixgbe_dev_start(struct rte_eth_dev *dev);
 static void ixgbe_dev_stop(struct rte_eth_dev *dev);
@@ -387,6 +388,7 @@ static int ixgbe_dev_udp_tunnel_port_add(struct rte_eth_dev 
*dev,
 struct rte_eth_udp_tunnel *udp_tunnel);
 static int ixgbe_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
 struct rte_eth_udp_tunnel *udp_tunnel);
+static int ixgbe_filter_restore(struct rte_eth_dev *dev);
 
 /*
  * Define VF Stats MACRO for Non "cleared on read" register
@@ -1336,6 +1338,27 @@ eth_ixgbe_dev_uninit(struct rte_eth_dev *eth_dev)
/* remove all the L2 tunnel filters & hash */
ixgbe_l2_tn_filter_uninit(eth_dev);
 
+   /* Remove all ntuple filters of the device */
+   ixgbe_ntuple_filter_uninit(eth_dev);
+
+   return 0;
+}
+
+static int ixgbe_ntuple_filter_uninit(struct rte_eth_dev *eth_dev)
+{
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(eth_dev->data->dev_private);
+   struct ixgbe_5tuple_filter *p_5tuple;
+
+   while ((p_5tuple = TAILQ_FIRST(&filter_info->fivetuple_list))) {
+   TAILQ_REMOVE(&filter_info->fivetuple_list,
+p_5tuple,
+entries);
+   rte_free(p_5tuple);
+   }
+   memset(filter_info->fivetuple_mask, 0,
+  sizeof(uint32_t) * IXGBE_5TUPLE_ARRAY_SIZE);
+
return 0;
 }
 
@@ -2504,6 +2527,7 @@ ixgbe_dev_start(struct rte_eth_dev *dev)
 
/* resume enabled intr since hw reset */
ixgbe_enable_intr(dev);
+   ixgbe_filter_restore(dev);
 
return 0;
 
@@ -2524,9 +2548,6 @@ ixgbe_dev_stop(struct rte_eth_dev *dev)
IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
struct ixgbe_vf_info *vfinfo =
*IXGBE_DEV_PRIVATE_TO_P_VFDATA(dev->data->dev_private);
-   struct ixgbe_filter_info *filter_info =
-   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
-   struct ixgbe_5tuple_filter *p_5tuple, *p_5tuple_next;
struct rte_pci_device *pci_dev = IXGBE_DEV_TO_PCI(dev);
struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
int vf;
@@ -2564,17 +2585,6 @@ ixgbe_dev_stop(struct rte_eth_dev *dev)
memset(&link, 0, sizeof(link));
rte_ixgbe_dev_atomic_write_link_status(dev, &link);
 
-   /* Remove all ntuple filters of the device */
-   for (p_5tuple = TAILQ_FIRST(&filter_info->fivetuple_list);
-p_5tuple != NULL; p_5tuple = p_5tuple_next) {
-   p_5tuple_next = TAILQ_NEXT(p_5tuple, entries);
-   TAILQ_REMOVE(&filter_info->fivetuple_list,
-p_5tuple, entries);
-   rte_free(p_5tuple);
-   }
-   memset(filter_info->fivetuple_mask, 0,
-   sizeof(uint32_t) * IXGBE_5TUPLE_ARRAY_SIZE);
-
if (!rte_intr_allow_others(intr_handle))
/* resume to the default handler */
rte_intr_callback_register(intr_handle,
@@ -5836,6 +5846,52 @@ convert_protocol_type(uint8_t protocol_value)
return IXGBE_FILTER_PROTOCOL_NONE;
 }
 
+/* inject a 5-tuple filter to HW */
+static inline void
+ixgbe_inject_5tuple_filter(struct rte_eth_dev *dev,
+  struct ixgbe_5tuple_filter *filter)
+{
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   int i;
+   uint32_t ftqf, sdpqf;
+   uint32_t l34timir = 0;
+   uint8_t mask = 0xff;
+
+   i = filter->index;
+
+   sdpqf = (uint32_t)(filter->filter_info.dst_port <<
+   IXGBE_SDPQF_DSTPORT_SHIFT);
+   sdpqf = sdpqf | (filter->filter_info.src_port & IXGBE_SDPQF_SRCPORT);
+
+   ftqf = (uint32_t)(filter->filter_info.proto &
+   IXGBE_FTQF_PROTOCOL_MASK);
+   ftqf |= (uint32_t)((filter->filter_info.priority &
+   IXGBE_FTQF_PRIORITY_MASK) << IXGBE_FTQF_PRIORITY_SHIFT);
+   if (filter->filter_info.src_ip_mask == 0) /* 0 means compare. */
+   mask &= IXGBE_FTQF_SOURCE_ADDR_MASK;
+

[dpdk-dev] [PATCH v6 05/18] net/ixgbe: restore ether type filter

[Wei Zhao]

Add support for restoring ether type filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 79 
 drivers/net/ixgbe/ixgbe_ethdev.h | 57 -
 drivers/net/ixgbe/ixgbe_pf.c | 25 -
 3 files changed, 104 insertions(+), 57 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 1630e65..6c46354 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -6259,47 +6259,6 @@ ixgbe_ntuple_filter_handle(struct rte_eth_dev *dev,
return ret;
 }
 
-static inline int
-ixgbe_ethertype_filter_lookup(struct ixgbe_filter_info *filter_info,
-   uint16_t ethertype)
-{
-   int i;
-
-   for (i = 0; i < IXGBE_MAX_ETQF_FILTERS; i++) {
-   if (filter_info->ethertype_filters[i] == ethertype &&
-   (filter_info->ethertype_mask & (1 << i)))
-   return i;
-   }
-   return -1;
-}
-
-static inline int
-ixgbe_ethertype_filter_insert(struct ixgbe_filter_info *filter_info,
-   uint16_t ethertype)
-{
-   int i;
-
-   for (i = 0; i < IXGBE_MAX_ETQF_FILTERS; i++) {
-   if (!(filter_info->ethertype_mask & (1 << i))) {
-   filter_info->ethertype_mask |= 1 << i;
-   filter_info->ethertype_filters[i] = ethertype;
-   return i;
-   }
-   }
-   return -1;
-}
-
-static inline int
-ixgbe_ethertype_filter_remove(struct ixgbe_filter_info *filter_info,
-   uint8_t idx)
-{
-   if (idx >= IXGBE_MAX_ETQF_FILTERS)
-   return -1;
-   filter_info->ethertype_mask &= ~(1 << idx);
-   filter_info->ethertype_filters[idx] = 0;
-   return idx;
-}
-
 static int
 ixgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
struct rte_eth_ethertype_filter *filter,
@@ -6311,6 +6270,7 @@ ixgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
uint32_t etqf = 0;
uint32_t etqs = 0;
int ret;
+   struct ixgbe_ethertype_filter ethertype_filter;
 
if (filter->queue >= IXGBE_MAX_RX_QUEUE_NUM)
return -EINVAL;
@@ -6344,18 +6304,22 @@ ixgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
}
 
if (add) {
-   ret = ixgbe_ethertype_filter_insert(filter_info,
-   filter->ether_type);
-   if (ret < 0) {
-   PMD_DRV_LOG(ERR, "ethertype filters are full.");
-   return -ENOSYS;
-   }
etqf = IXGBE_ETQF_FILTER_EN;
etqf |= (uint32_t)filter->ether_type;
etqs |= (uint32_t)((filter->queue <<
IXGBE_ETQS_RX_QUEUE_SHIFT) &
IXGBE_ETQS_RX_QUEUE);
etqs |= IXGBE_ETQS_QUEUE_EN;
+
+   ethertype_filter.ethertype = filter->ether_type;
+   ethertype_filter.etqf = etqf;
+   ethertype_filter.etqs = etqs;
+   ret = ixgbe_ethertype_filter_insert(filter_info,
+   ðertype_filter);
+   if (ret < 0) {
+   PMD_DRV_LOG(ERR, "ethertype filters are full.");
+   return -ENOSPC;
+   }
} else {
ret = ixgbe_ethertype_filter_remove(filter_info, (uint8_t)ret);
if (ret < 0)
@@ -7961,10 +7925,31 @@ ixgbe_ntuple_filter_restore(struct rte_eth_dev *dev)
}
 }
 
+/* restore ethernet type filter */
+static inline void
+ixgbe_ethertype_filter_restore(struct rte_eth_dev *dev)
+{
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
+   int i;
+
+   for (i = 0; i < IXGBE_MAX_ETQF_FILTERS; i++) {
+   if (filter_info->ethertype_mask & (1 << i)) {
+   IXGBE_WRITE_REG(hw, IXGBE_ETQF(i),
+   filter_info->ethertype_filters[i].etqf);
+   IXGBE_WRITE_REG(hw, IXGBE_ETQS(i),
+   filter_info->ethertype_filters[i].etqs);
+   IXGBE_WRITE_FLUSH(hw);
+   }
+   }
+}
+
 static int
 ixgbe_filter_restore(struct rte_eth_dev *dev)
 {
ixgbe_ntuple_filter_restore(dev);
+   ixgbe_ethertype_filter_restore(dev);
 
return 0;
 }
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index b2cf789..36aae01 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -270,13 +270,19 @@ struct ixgbe_5tuple_filter {
(RTE_ALIGN(IXGBE_MAX_FTQF_FILTERS, (sizeof(uint32_t) * NBBY)) / \
 (sizeof(uint

[dpdk-dev] [PATCH v6 03/18] net/ixgbe: store L2 tunnel filter

[Wei Zhao]

Add support for storing L2 tunnel filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 172 ++-
 drivers/net/ixgbe/ixgbe_ethdev.h |  24 ++
 2 files changed, 193 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 9796c4f..e63b635 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -168,6 +168,8 @@ static int eth_ixgbe_dev_init(struct rte_eth_dev *eth_dev);
 static int eth_ixgbe_dev_uninit(struct rte_eth_dev *eth_dev);
 static int ixgbe_fdir_filter_init(struct rte_eth_dev *eth_dev);
 static int ixgbe_fdir_filter_uninit(struct rte_eth_dev *eth_dev);
+static int ixgbe_l2_tn_filter_init(struct rte_eth_dev *eth_dev);
+static int ixgbe_l2_tn_filter_uninit(struct rte_eth_dev *eth_dev);
 static int  ixgbe_dev_configure(struct rte_eth_dev *dev);
 static int  ixgbe_dev_start(struct rte_eth_dev *dev);
 static void ixgbe_dev_stop(struct rte_eth_dev *dev);
@@ -1285,6 +1287,8 @@ eth_ixgbe_dev_init(struct rte_eth_dev *eth_dev)
/* initialize flow director filter list & hash */
ixgbe_fdir_filter_init(eth_dev);
 
+   /* initialize l2 tunnel filter list & hash */
+   ixgbe_l2_tn_filter_init(eth_dev);
return 0;
 }
 
@@ -1329,6 +1333,9 @@ eth_ixgbe_dev_uninit(struct rte_eth_dev *eth_dev)
/* remove all the fdir filters & hash */
ixgbe_fdir_filter_uninit(eth_dev);
 
+   /* remove all the L2 tunnel filters & hash */
+   ixgbe_l2_tn_filter_uninit(eth_dev);
+
return 0;
 }
 
@@ -1353,6 +1360,27 @@ static int ixgbe_fdir_filter_uninit(struct rte_eth_dev 
*eth_dev)
return 0;
 }
 
+static int ixgbe_l2_tn_filter_uninit(struct rte_eth_dev *eth_dev)
+{
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(eth_dev->data->dev_private);
+   struct ixgbe_l2_tn_filter *l2_tn_filter;
+
+   if (l2_tn_info->hash_map)
+   rte_free(l2_tn_info->hash_map);
+   if (l2_tn_info->hash_handle)
+   rte_hash_free(l2_tn_info->hash_handle);
+
+   while ((l2_tn_filter = TAILQ_FIRST(&l2_tn_info->l2_tn_list))) {
+   TAILQ_REMOVE(&l2_tn_info->l2_tn_list,
+l2_tn_filter,
+entries);
+   rte_free(l2_tn_filter);
+   }
+
+   return 0;
+}
+
 static int ixgbe_fdir_filter_init(struct rte_eth_dev *eth_dev)
 {
struct ixgbe_hw_fdir_info *fdir_info =
@@ -1384,6 +1412,40 @@ static int ixgbe_fdir_filter_init(struct rte_eth_dev 
*eth_dev)
 "Failed to allocate memory for fdir hash map!");
return -ENOMEM;
}
+   return 0;
+}
+
+static int ixgbe_l2_tn_filter_init(struct rte_eth_dev *eth_dev)
+{
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(eth_dev->data->dev_private);
+   char l2_tn_hash_name[RTE_HASH_NAMESIZE];
+   struct rte_hash_parameters l2_tn_hash_params = {
+   .name = l2_tn_hash_name,
+   .entries = IXGBE_MAX_L2_TN_FILTER_NUM,
+   .key_len = sizeof(struct ixgbe_l2_tn_key),
+   .hash_func = rte_hash_crc,
+   .hash_func_init_val = 0,
+   .socket_id = rte_socket_id(),
+   };
+
+   TAILQ_INIT(&l2_tn_info->l2_tn_list);
+   snprintf(l2_tn_hash_name, RTE_HASH_NAMESIZE,
+"l2_tn_%s", eth_dev->data->name);
+   l2_tn_info->hash_handle = rte_hash_create(&l2_tn_hash_params);
+   if (!l2_tn_info->hash_handle) {
+   PMD_INIT_LOG(ERR, "Failed to create L2 TN hash table!");
+   return -EINVAL;
+   }
+   l2_tn_info->hash_map = rte_zmalloc("ixgbe",
+  sizeof(struct ixgbe_l2_tn_filter *) *
+  IXGBE_MAX_L2_TN_FILTER_NUM,
+  0);
+   if (!l2_tn_info->hash_map) {
+   PMD_INIT_LOG(ERR,
+   "Failed to allocate memory for L2 TN hash map!");
+   return -ENOMEM;
+   }
 
return 0;
 }
@@ -7210,12 +7272,104 @@ ixgbe_e_tag_filter_add(struct rte_eth_dev *dev,
return -EINVAL;
 }
 
+static inline struct ixgbe_l2_tn_filter *
+ixgbe_l2_tn_filter_lookup(struct ixgbe_l2_tn_info *l2_tn_info,
+ struct ixgbe_l2_tn_key *key)
+{
+   int ret;
+
+   ret = rte_hash_lookup(l2_tn_info->hash_handle, (const void *)key);
+   if (ret < 0)
+   return NULL;
+
+   return l2_tn_info->hash_map[ret];
+}
+
+static inline int
+ixgbe_insert_l2_tn_filter(struct ixgbe_l2_tn_info *l2_tn_info,
+ struct ixgbe_l2_tn_filter *l2_tn_filter)
+{
+   int ret;
+
+   ret = rte_hash_add_key(l2_tn_info->hash_handle,
+  &l2_tn_filter->key);
+
+   if (ret < 0) {
+   PMD_DRV_LOG(ERR,
+   

[dpdk-dev] [PATCH v6 07/18] net/ixgbe: restore flow director filter

[Wei Zhao]

Add support for storing flow director filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |  1 +
 drivers/net/ixgbe/ixgbe_ethdev.h |  1 +
 drivers/net/ixgbe/ixgbe_fdir.c   | 35 +++
 3 files changed, 37 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 6cd5975..f48e30c 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -7968,6 +7968,7 @@ ixgbe_filter_restore(struct rte_eth_dev *dev)
ixgbe_ntuple_filter_restore(dev);
ixgbe_ethertype_filter_restore(dev);
ixgbe_syn_filter_restore(dev);
+   ixgbe_fdir_filter_restore(dev);
 
return 0;
 }
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index 36aae01..4f281e8 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -497,6 +497,7 @@ uint32_t ixgbe_convert_vm_rx_mask_to_val(uint16_t rx_mask, 
uint32_t orig_val);
 
 int ixgbe_fdir_ctrl_func(struct rte_eth_dev *dev,
enum rte_filter_op filter_op, void *arg);
+void ixgbe_fdir_filter_restore(struct rte_eth_dev *dev);
 
 static inline int
 ixgbe_ethertype_filter_lookup(struct ixgbe_filter_info *filter_info,
diff --git a/drivers/net/ixgbe/ixgbe_fdir.c b/drivers/net/ixgbe/ixgbe_fdir.c
index 8bf5705..627c51a 100644
--- a/drivers/net/ixgbe/ixgbe_fdir.c
+++ b/drivers/net/ixgbe/ixgbe_fdir.c
@@ -1479,3 +1479,38 @@ ixgbe_fdir_ctrl_func(struct rte_eth_dev *dev,
}
return ret;
 }
+
+/* restore flow director filter */
+void
+ixgbe_fdir_filter_restore(struct rte_eth_dev *dev)
+{
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_hw_fdir_info *fdir_info =
+   IXGBE_DEV_PRIVATE_TO_FDIR_INFO(dev->data->dev_private);
+   struct ixgbe_fdir_filter *node;
+   bool is_perfect = FALSE;
+   enum rte_fdir_mode fdir_mode = dev->data->dev_conf.fdir_conf.mode;
+
+   if (fdir_mode >= RTE_FDIR_MODE_PERFECT &&
+   fdir_mode <= RTE_FDIR_MODE_PERFECT_TUNNEL)
+   is_perfect = TRUE;
+
+   if (is_perfect) {
+   TAILQ_FOREACH(node, &fdir_info->fdir_list, entries) {
+   (void)fdir_write_perfect_filter_82599(hw,
+ &node->ixgbe_fdir,
+ node->queue,
+ node->fdirflags,
+ node->fdirhash,
+ fdir_mode);
+   }
+   } else {
+   TAILQ_FOREACH(node, &fdir_info->fdir_list, entries) {
+   (void)fdir_add_signature_filter_82599(hw,
+ &node->ixgbe_fdir,
+ node->queue,
+ node->fdirflags,
+ node->fdirhash);
+   }
+   }
+}
-- 
2.5.5

[dpdk-dev] [PATCH v6 08/18] net/ixgbe: restore L2 tunnel filter

[Wei Zhao]

Add support for restoring L2 tunnel filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 69 ++--
 1 file changed, 46 insertions(+), 23 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index f48e30c..0ad613b 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -7324,7 +7324,8 @@ ixgbe_remove_l2_tn_filter(struct ixgbe_l2_tn_info 
*l2_tn_info,
 /* Add l2 tunnel filter */
 static int
 ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
-  struct rte_eth_l2_tunnel_conf *l2_tunnel)
+  struct rte_eth_l2_tunnel_conf *l2_tunnel,
+  bool restore)
 {
int ret;
struct ixgbe_l2_tn_info *l2_tn_info =
@@ -7332,30 +7333,33 @@ ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
struct ixgbe_l2_tn_key key;
struct ixgbe_l2_tn_filter *node;
 
-   key.l2_tn_type = l2_tunnel->l2_tunnel_type;
-   key.tn_id = l2_tunnel->tunnel_id;
+   if (!restore) {
+   key.l2_tn_type = l2_tunnel->l2_tunnel_type;
+   key.tn_id = l2_tunnel->tunnel_id;
 
-   node = ixgbe_l2_tn_filter_lookup(l2_tn_info, &key);
+   node = ixgbe_l2_tn_filter_lookup(l2_tn_info, &key);
 
-   if (node) {
-   PMD_DRV_LOG(ERR, "The L2 tunnel filter already exists!");
-   return -EINVAL;
-   }
+   if (node) {
+   PMD_DRV_LOG(ERR,
+   "The L2 tunnel filter already exists!");
+   return -EINVAL;
+   }
 
-   node = rte_zmalloc("ixgbe_l2_tn",
-  sizeof(struct ixgbe_l2_tn_filter),
-  0);
-   if (!node)
-   return -ENOMEM;
+   node = rte_zmalloc("ixgbe_l2_tn",
+  sizeof(struct ixgbe_l2_tn_filter),
+  0);
+   if (!node)
+   return -ENOMEM;
 
-   (void)rte_memcpy(&node->key,
-&key,
-sizeof(struct ixgbe_l2_tn_key));
-   node->pool = l2_tunnel->pool;
-   ret = ixgbe_insert_l2_tn_filter(l2_tn_info, node);
-   if (ret < 0) {
-   rte_free(node);
-   return ret;
+   (void)rte_memcpy(&node->key,
+&key,
+sizeof(struct ixgbe_l2_tn_key));
+   node->pool = l2_tunnel->pool;
+   ret = ixgbe_insert_l2_tn_filter(l2_tn_info, node);
+   if (ret < 0) {
+   rte_free(node);
+   return ret;
+   }
}
 
switch (l2_tunnel->l2_tunnel_type) {
@@ -7368,7 +7372,7 @@ ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
break;
}
 
-   if (ret < 0)
+   if ((!restore) && (ret < 0))
(void)ixgbe_remove_l2_tn_filter(l2_tn_info, &key);
 
return ret;
@@ -7429,7 +7433,8 @@ ixgbe_dev_l2_tunnel_filter_handle(struct rte_eth_dev *dev,
case RTE_ETH_FILTER_ADD:
ret = ixgbe_dev_l2_tunnel_filter_add
(dev,
-(struct rte_eth_l2_tunnel_conf *)arg);
+(struct rte_eth_l2_tunnel_conf *)arg,
+FALSE);
break;
case RTE_ETH_FILTER_DELETE:
ret = ixgbe_dev_l2_tunnel_filter_del
@@ -7962,6 +7967,23 @@ ixgbe_syn_filter_restore(struct rte_eth_dev *dev)
}
 }
 
+/* restore L2 tunnel filter */
+static inline void
+ixgbe_l2_tn_filter_restore(struct rte_eth_dev *dev)
+{
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
+   struct ixgbe_l2_tn_filter *node;
+   struct rte_eth_l2_tunnel_conf l2_tn_conf;
+
+   TAILQ_FOREACH(node, &l2_tn_info->l2_tn_list, entries) {
+   l2_tn_conf.l2_tunnel_type = node->key.l2_tn_type;
+   l2_tn_conf.tunnel_id  = node->key.tn_id;
+   l2_tn_conf.pool   = node->pool;
+   (void)ixgbe_dev_l2_tunnel_filter_add(dev, &l2_tn_conf, TRUE);
+   }
+}
+
 static int
 ixgbe_filter_restore(struct rte_eth_dev *dev)
 {
@@ -7969,6 +7991,7 @@ ixgbe_filter_restore(struct rte_eth_dev *dev)
ixgbe_ethertype_filter_restore(dev);
ixgbe_syn_filter_restore(dev);
ixgbe_fdir_filter_restore(dev);
+   ixgbe_l2_tn_filter_restore(dev);
 
return 0;
 }
-- 
2.5.5

[dpdk-dev] [PATCH v6 10/18] net/ixgbe: flush all the filters

[Wei Zhao]

Add support for flush all the filters in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/Makefile   |   2 +
 drivers/net/ixgbe/ixgbe_ethdev.c |  79 ++-
 drivers/net/ixgbe/ixgbe_ethdev.h |  16 ++
 drivers/net/ixgbe/ixgbe_fdir.c   |  24 
 drivers/net/ixgbe/ixgbe_flow.c   | 115 +++
 drivers/net/ixgbe/ixgbe_pf.c |   1 +
 6 files changed, 236 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ixgbe/ixgbe_flow.c

diff --git a/drivers/net/ixgbe/Makefile b/drivers/net/ixgbe/Makefile
index a3e6a52..38b9fbd 100644
--- a/drivers/net/ixgbe/Makefile
+++ b/drivers/net/ixgbe/Makefile
@@ -109,6 +109,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe_rxtx.c
 SRCS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe_ethdev.c
 SRCS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe_fdir.c
 SRCS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe_pf.c
+SRCS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe_flow.c
 ifeq ($(CONFIG_RTE_ARCH_ARM64),y)
 SRCS-$(CONFIG_RTE_IXGBE_INC_VECTOR) += ixgbe_rxtx_vec_neon.c
 else
@@ -127,5 +128,6 @@ SYMLINK-$(CONFIG_RTE_LIBRTE_IXGBE_PMD)-include := 
rte_pmd_ixgbe.h
 DEPDIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += lib/librte_eal lib/librte_ether
 DEPDIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += lib/librte_mempool lib/librte_mbuf
 DEPDIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += lib/librte_net
+DEPDIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += lib/librte_hash
 
 include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 3059c64..2a67462 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -6319,6 +6319,7 @@ ixgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
ethertype_filter.ethertype = filter->ether_type;
ethertype_filter.etqf = etqf;
ethertype_filter.etqs = etqs;
+   ethertype_filter.conf = FALSE;
ret = ixgbe_ethertype_filter_insert(filter_info,
ðertype_filter);
if (ret < 0) {
@@ -6420,7 +6421,7 @@ ixgbe_dev_filter_ctrl(struct rte_eth_dev *dev,
 enum rte_filter_op filter_op,
 void *arg)
 {
-   int ret = -EINVAL;
+   int ret = 0;
 
switch (filter_type) {
case RTE_ETH_FILTER_NTUPLE:
@@ -6438,9 +6439,15 @@ ixgbe_dev_filter_ctrl(struct rte_eth_dev *dev,
case RTE_ETH_FILTER_L2_TUNNEL:
ret = ixgbe_dev_l2_tunnel_filter_handle(dev, filter_op, arg);
break;
+   case RTE_ETH_FILTER_GENERIC:
+   if (filter_op != RTE_ETH_FILTER_GET)
+   return -EINVAL;
+   *(const void **)arg = &ixgbe_flow_ops;
+   break;
default:
PMD_DRV_LOG(WARNING, "Filter type (%d) not supported",
filter_type);
+   ret = -EINVAL;
break;
}
 
@@ -8032,6 +8039,76 @@ ixgbe_l2_tunnel_conf(struct rte_eth_dev *dev)
(void)ixgbe_update_e_tag_eth_type(hw, l2_tn_info->e_tag_ether_type);
 }
 
+/* remove all the n-tuple filters */
+void
+ixgbe_clear_all_ntuple_filter(struct rte_eth_dev *dev)
+{
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
+   struct ixgbe_5tuple_filter *p_5tuple;
+
+   while ((p_5tuple = TAILQ_FIRST(&filter_info->fivetuple_list)))
+   ixgbe_remove_5tuple_filter(dev, p_5tuple);
+}
+
+/* remove all the ether type filters */
+void
+ixgbe_clear_all_ethertype_filter(struct rte_eth_dev *dev)
+{
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
+   int i;
+
+   for (i = 0; i < IXGBE_MAX_ETQF_FILTERS; i++) {
+   if (filter_info->ethertype_mask & (1 << i) &&
+   !filter_info->ethertype_filters[i].conf) {
+   (void)ixgbe_ethertype_filter_remove(filter_info,
+   (uint8_t)i);
+   IXGBE_WRITE_REG(hw, IXGBE_ETQF(i), 0);
+   IXGBE_WRITE_REG(hw, IXGBE_ETQS(i), 0);
+   IXGBE_WRITE_FLUSH(hw);
+   }
+   }
+}
+
+/* remove the SYN filter */
+void
+ixgbe_clear_syn_filter(struct rte_eth_dev *dev)
+{
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_filter_info *filter_info =
+   IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
+
+   if (filter_info->syn_info & IXGBE_SYN_FILTER_ENABLE) {
+   filter_info->syn_info = 0;
+
+   IXGBE_WRITE_REG(hw, IXGBE_SYNQF, 0);
+   IXGBE_WRITE_FLUSH(hw);
+   }
+}
+
+/* remove all the L2 tunnel filters */
+int ixg

[dpdk-dev] [PATCH v6 09/18] net/ixgbe: store and restore L2 tunnel configuration

[Wei Zhao]

Add support for store and restore L2 tunnel filter in SW.

Signed-off-by: Wenzhuo Lu 
Signed-off-by: Wei Zhao 
---
 drivers/net/ixgbe/ixgbe_ethdev.c | 36 
 drivers/net/ixgbe/ixgbe_ethdev.h |  3 +++
 2 files changed, 39 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 0ad613b..3059c64 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -389,6 +389,7 @@ static int ixgbe_dev_udp_tunnel_port_add(struct rte_eth_dev 
*dev,
 static int ixgbe_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
 struct rte_eth_udp_tunnel *udp_tunnel);
 static int ixgbe_filter_restore(struct rte_eth_dev *dev);
+static void ixgbe_l2_tunnel_conf(struct rte_eth_dev *dev);
 
 /*
  * Define VF Stats MACRO for Non "cleared on read" register
@@ -1469,6 +1470,9 @@ static int ixgbe_l2_tn_filter_init(struct rte_eth_dev 
*eth_dev)
"Failed to allocate memory for L2 TN hash map!");
return -ENOMEM;
}
+   l2_tn_info->e_tag_en = FALSE;
+   l2_tn_info->e_tag_fwd_en = FALSE;
+   l2_tn_info->e_tag_ether_type = DEFAULT_ETAG_ETYPE;
 
return 0;
 }
@@ -2527,6 +2531,7 @@ ixgbe_dev_start(struct rte_eth_dev *dev)
 
/* resume enabled intr since hw reset */
ixgbe_enable_intr(dev);
+   ixgbe_l2_tunnel_conf(dev);
ixgbe_filter_restore(dev);
 
return 0;
@@ -7083,12 +7088,15 @@ ixgbe_dev_l2_tunnel_eth_type_conf(struct rte_eth_dev 
*dev,
 {
int ret = 0;
struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
 
if (l2_tunnel == NULL)
return -EINVAL;
 
switch (l2_tunnel->l2_tunnel_type) {
case RTE_L2_TUNNEL_TYPE_E_TAG:
+   l2_tn_info->e_tag_ether_type = l2_tunnel->ether_type;
ret = ixgbe_update_e_tag_eth_type(hw, l2_tunnel->ether_type);
break;
default:
@@ -7127,9 +7135,12 @@ ixgbe_dev_l2_tunnel_enable(struct rte_eth_dev *dev,
 {
int ret = 0;
struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
 
switch (l2_tunnel_type) {
case RTE_L2_TUNNEL_TYPE_E_TAG:
+   l2_tn_info->e_tag_en = TRUE;
ret = ixgbe_e_tag_enable(hw);
break;
default:
@@ -7168,9 +7179,12 @@ ixgbe_dev_l2_tunnel_disable(struct rte_eth_dev *dev,
 {
int ret = 0;
struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
 
switch (l2_tunnel_type) {
case RTE_L2_TUNNEL_TYPE_E_TAG:
+   l2_tn_info->e_tag_en = FALSE;
ret = ixgbe_e_tag_disable(hw);
break;
default:
@@ -7477,10 +7491,13 @@ ixgbe_dev_l2_tunnel_forwarding_enable
(struct rte_eth_dev *dev,
 enum rte_eth_tunnel_type l2_tunnel_type)
 {
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
int ret = 0;
 
switch (l2_tunnel_type) {
case RTE_L2_TUNNEL_TYPE_E_TAG:
+   l2_tn_info->e_tag_fwd_en = TRUE;
ret = ixgbe_e_tag_forwarding_en_dis(dev, 1);
break;
default:
@@ -7498,10 +7515,13 @@ ixgbe_dev_l2_tunnel_forwarding_disable
(struct rte_eth_dev *dev,
 enum rte_eth_tunnel_type l2_tunnel_type)
 {
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
int ret = 0;
 
switch (l2_tunnel_type) {
case RTE_L2_TUNNEL_TYPE_E_TAG:
+   l2_tn_info->e_tag_fwd_en = FALSE;
ret = ixgbe_e_tag_forwarding_en_dis(dev, 0);
break;
default:
@@ -7996,6 +8016,22 @@ ixgbe_filter_restore(struct rte_eth_dev *dev)
return 0;
 }
 
+static void
+ixgbe_l2_tunnel_conf(struct rte_eth_dev *dev)
+{
+   struct ixgbe_l2_tn_info *l2_tn_info =
+   IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
+   struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+   if (l2_tn_info->e_tag_en)
+   (void)ixgbe_e_tag_enable(hw);
+
+   if (l2_tn_info->e_tag_fwd_en)
+   (void)ixgbe_e_tag_forwarding_en_dis(dev, 1);
+
+   (void)ixgbe_update_e_tag_eth_type(hw, l2_tn_info->e_tag_ether_type);
+}
+
 RTE_PMD_REGISTER_PCI(net_ixgbe, rte_ixgbe_pmd.pci_drv);
 RTE_PMD_REGISTER_PCI_TABLE(net_ixgbe, pci_id_ixgbe_map);
 RTE_PMD_REGISTER_KMOD_DEP(net_ixgbe, "* igb_uio | uio_pci_generic | vfio");
diff --git a/drivers/n

[dpdk-dev] [PATCH v6 11/18] net/ixgbe: parse n-tuple filter

[Wei Zhao]

Add rule validate function and check if the rule is a n-tuple rule,
and get the n-tuple info.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_flow.c | 430 -
 1 file changed, 429 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
index 1499391..417728b 100644
--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c
@@ -77,15 +77,443 @@
 
 static int ixgbe_flow_flush(struct rte_eth_dev *dev,
struct rte_flow_error *error);
+static int
+cons_parse_ntuple_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_ntuple_filter *filter,
+   struct rte_flow_error *error);
+static int
+ixgbe_parse_ntuple_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_ntuple_filter *filter,
+   struct rte_flow_error *error);
+static int
+ixgbe_flow_validate(__rte_unused struct rte_eth_dev *dev,
+   const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_flow_error *error);
 
 const struct rte_flow_ops ixgbe_flow_ops = {
-   NULL,
+   ixgbe_flow_validate,
NULL,
NULL,
ixgbe_flow_flush,
NULL,
 };
 
+#define IXGBE_MIN_N_TUPLE_PRIO 1
+#define IXGBE_MAX_N_TUPLE_PRIO 7
+#define NEXT_ITEM_OF_PATTERN(item, pattern, index)\
+   do {\
+   item = pattern + index;\
+   while (item->type == RTE_FLOW_ITEM_TYPE_VOID) {\
+   index++;\
+   item = pattern + index; \
+   }   \
+   } while (0)
+
+#define NEXT_ITEM_OF_ACTION(act, actions, index)\
+   do {\
+   act = actions + index;  \
+   while (act->type == RTE_FLOW_ACTION_TYPE_VOID) {\
+   index++;\
+   act = actions + index;  \
+   }   \
+   } while (0)
+
+/**
+ * Please aware there's an asumption for all the parsers.
+ * rte_flow_item is using big endian, rte_flow_attr and
+ * rte_flow_action are using CPU order.
+ * Because the pattern is used to describe the packets,
+ * normally the packets should use network order.
+ */
+
+/**
+ * Parse the rule to see if it is a n-tuple rule.
+ * And get the n-tuple filter info BTW.
+ * pattern:
+ * The first not void item can be ETH or IPV4.
+ * The second not void item must be IPV4 if the first one is ETH.
+ * The third not void item must be UDP or TCP.
+ * The next not void item must be END.
+ * action:
+ * The first not void action should be QUEUE.
+ * The next not void action should be END.
+ * pattern example:
+ * ITEMSpecMask
+ * ETH NULLNULL
+ * IPV4src_addr 192.168.1.20   0x
+ * dst_addr 192.167.3.50   0x
+ * next_proto_id   17  0xFF
+ * UDP/TCP src_port80  0x
+ * dst_port80  0x
+ * END
+ * other members in mask and spec should set to 0x00.
+ * item->last should be NULL.
+ */
+static int
+cons_parse_ntuple_filter(const struct rte_flow_attr *attr,
+const struct rte_flow_item pattern[],
+const struct rte_flow_action actions[],
+struct rte_eth_ntuple_filter *filter,
+struct rte_flow_error *error)
+{
+   const struct rte_flow_item *item;
+   const struct rte_flow_action *act;
+   const struct rte_flow_item_ipv4 *ipv4_spec;
+   const struct rte_flow_item_ipv4 *ipv4_mask;
+   const struct rte_flow_item_tcp *tcp_spec;
+   const struct rte_flow_item_tcp *tcp_mask;
+   const struct rte_flow_item_udp *udp_spec;
+   const struct rte_flow_item_udp *udp_mask;
+   uint32_t index;
+
+   if (!pattern) {
+   rte_flow_error_set(error,
+   EINVAL, RTE_FLOW_ERROR_TYPE_ITEM_NUM,
+   NULL, "NULL pattern.");
+   return -rte_errno;
+   }
+
+   if (!actions) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ACTION_NUM,
+   

[dpdk-dev] [PATCH v6 12/18] net/ixgbe: parse ethertype filter

[Wei Zhao]

check if the rule is a ethertype rule, and get the ethertype info.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_flow.c | 284 +
 1 file changed, 284 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
index 417728b..2e2cafd 100644
--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c
@@ -90,6 +90,18 @@ ixgbe_parse_ntuple_filter(const struct rte_flow_attr *attr,
struct rte_eth_ntuple_filter *filter,
struct rte_flow_error *error);
 static int
+cons_parse_ethertype_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item *pattern,
+   const struct rte_flow_action *actions,
+   struct rte_eth_ethertype_filter *filter,
+   struct rte_flow_error *error);
+static int
+ixgbe_parse_ethertype_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_ethertype_filter *filter,
+   struct rte_flow_error *error);
+static int
 ixgbe_flow_validate(__rte_unused struct rte_eth_dev *dev,
const struct rte_flow_attr *attr,
const struct rte_flow_item pattern[],
@@ -491,6 +503,271 @@ ixgbe_parse_ntuple_filter(const struct rte_flow_attr 
*attr,
 }
 
 /**
+ * Parse the rule to see if it is a ethertype rule.
+ * And get the ethertype filter info BTW.
+ * pattern:
+ * The first not void item can be ETH.
+ * The next not void item must be END.
+ * action:
+ * The first not void action should be QUEUE.
+ * The next not void action should be END.
+ * pattern example:
+ * ITEMSpecMask
+ * ETH type0x0807  0x
+ * END
+ * other members in mask and spec should set to 0x00.
+ * item->last should be NULL.
+ */
+static int
+cons_parse_ethertype_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item *pattern,
+   const struct rte_flow_action *actions,
+   struct rte_eth_ethertype_filter *filter,
+   struct rte_flow_error *error)
+{
+   const struct rte_flow_item *item;
+   const struct rte_flow_action *act;
+   const struct rte_flow_item_eth *eth_spec;
+   const struct rte_flow_item_eth *eth_mask;
+   const struct rte_flow_action_queue *act_q;
+   uint32_t index;
+
+   if (!pattern) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM_NUM,
+   NULL, "NULL pattern.");
+   return -rte_errno;
+   }
+
+   if (!actions) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ACTION_NUM,
+   NULL, "NULL action.");
+   return -rte_errno;
+   }
+
+   if (!attr) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ATTR,
+  NULL, "NULL attribute.");
+   return -rte_errno;
+   }
+
+   /* Parse pattern */
+   index = 0;
+
+   /* The first non-void item should be MAC. */
+   item = pattern + index;
+   while (item->type == RTE_FLOW_ITEM_TYPE_VOID) {
+   index++;
+   item = pattern + index;
+   }
+   if (item->type != RTE_FLOW_ITEM_TYPE_ETH) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM,
+   item, "Not supported by ethertype filter");
+   return -rte_errno;
+   }
+
+   /*Not supported last point for range*/
+   if (item->last) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+   item, "Not supported last point for range");
+   return -rte_errno;
+   }
+
+   /* Get the MAC info. */
+   if (!item->spec || !item->mask) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM,
+   item, "Not supported by ethertype filter");
+   return -rte_errno;
+   }
+
+   eth_spec = (const struct rte_flow_item_eth *)item->spec;
+   eth_mask = (const struct rte_flow_item_eth *)item->mask;
+
+   /* Mask bits of source MAC address must be full of 0.
+* Mask bits of destination MAC address must be full
+* of 1 or full of 0.
+*/
+   if (!is_zero_ether_addr(ð_mask->src) ||
+   (!is_zero_ether_addr(ð_mask->dst) &&
+!is_broadcast_ether_add

[dpdk-dev] [PATCH v6 14/18] net/ixgbe: parse L2 tunnel filter

[Wei Zhao]

check if the rule is a L2 tunnel rule, and get the L2 tunnel info.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |   3 +-
 drivers/net/ixgbe/ixgbe_flow.c   | 216 +++
 lib/librte_ether/rte_flow.h  |  48 +
 3 files changed, 266 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 2a67462..14a88ee 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -8089,7 +8089,8 @@ ixgbe_clear_syn_filter(struct rte_eth_dev *dev)
 }
 
 /* remove all the L2 tunnel filters */
-int ixgbe_clear_all_l2_tn_filter(struct rte_eth_dev *dev)
+int
+ixgbe_clear_all_l2_tn_filter(struct rte_eth_dev *dev)
 {
struct ixgbe_l2_tn_info *l2_tn_info =
IXGBE_DEV_PRIVATE_TO_L2_TN_INFO(dev->data->dev_private);
diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
index 7557dfa..4006084 100644
--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c
@@ -114,6 +114,19 @@ ixgbe_parse_syn_filter(const struct rte_flow_attr *attr,
struct rte_eth_syn_filter *filter,
struct rte_flow_error *error);
 static int
+cons_parse_l2_tn_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_l2_tunnel_conf *filter,
+   struct rte_flow_error *error);
+static int
+ixgbe_validate_l2_tn_filter(struct rte_eth_dev *dev,
+   const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_l2_tunnel_conf *rule,
+   struct rte_flow_error *error);
+static int
 ixgbe_flow_validate(__rte_unused struct rte_eth_dev *dev,
const struct rte_flow_attr *attr,
const struct rte_flow_item pattern[],
@@ -1033,6 +1046,204 @@ ixgbe_parse_syn_filter(const struct rte_flow_attr *attr,
 }
 
 /**
+ * Parse the rule to see if it is a L2 tunnel rule.
+ * And get the L2 tunnel filter info BTW.
+ * Only support E-tag now.
+ * pattern:
+ * The first not void item can be E_TAG.
+ * The next not void item must be END.
+ * action:
+ * The first not void action should be QUEUE.
+ * The next not void action should be END.
+ * pattern example:
+ * ITEMSpecMask
+ * E_TAG   grp 0x1 0x3
+   e_cid_base  0x309   0xFFF
+ * END
+ * other members in mask and spec should set to 0x00.
+ * item->last should be NULL.
+ */
+static int
+cons_parse_l2_tn_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_l2_tunnel_conf *filter,
+   struct rte_flow_error *error)
+{
+   const struct rte_flow_item *item;
+   const struct rte_flow_item_e_tag *e_tag_spec;
+   const struct rte_flow_item_e_tag *e_tag_mask;
+   const struct rte_flow_action *act;
+   const struct rte_flow_action_queue *act_q;
+   uint32_t index;
+
+   if (!pattern) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM_NUM,
+   NULL, "NULL pattern.");
+   return -rte_errno;
+   }
+
+   if (!actions) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ACTION_NUM,
+  NULL, "NULL action.");
+   return -rte_errno;
+   }
+
+   if (!attr) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ATTR,
+  NULL, "NULL attribute.");
+   return -rte_errno;
+   }
+   /* parse pattern */
+   index = 0;
+
+   /* The first not void item should be e-tag. */
+   NEXT_ITEM_OF_PATTERN(item, pattern, index);
+   if (item->type != RTE_FLOW_ITEM_TYPE_E_TAG) {
+   memset(filter, 0, sizeof(struct rte_eth_l2_tunnel_conf));
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM,
+   item, "Not supported by L2 tunnel filter");
+   return -rte_errno;
+   }
+
+   if (!item->spec || !item->mask) {
+   memset(filter, 0, sizeof(struct rte_eth_l2_tunnel_conf));
+   rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+   item, "Not supported by L2 tunnel filter");
+   return -rte_errno;
+   }
+
+   /*Not supported last point for range*/
+   if (item->last) {
+   rte_flow_error_set(error, E

[dpdk-dev] [PATCH v6 13/18] net/ixgbe: parse TCP SYN filter

[Wei Zhao]

check if the rule is a TCP SYN rule, and get the SYN info.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_flow.c | 272 +
 1 file changed, 272 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
index 2e2cafd..7557dfa 100644
--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c
@@ -102,6 +102,18 @@ ixgbe_parse_ethertype_filter(const struct rte_flow_attr 
*attr,
struct rte_eth_ethertype_filter *filter,
struct rte_flow_error *error);
 static int
+cons_parse_syn_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_syn_filter *filter,
+   struct rte_flow_error *error);
+static int
+ixgbe_parse_syn_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_syn_filter *filter,
+   struct rte_flow_error *error);
+static int
 ixgbe_flow_validate(__rte_unused struct rte_eth_dev *dev,
const struct rte_flow_attr *attr,
const struct rte_flow_item pattern[],
@@ -768,6 +780,259 @@ ixgbe_parse_ethertype_filter(const struct rte_flow_attr 
*attr,
 }
 
 /**
+ * Parse the rule to see if it is a TCP SYN rule.
+ * And get the TCP SYN filter info BTW.
+ * pattern:
+ * The first not void item must be ETH.
+ * The second not void item must be IPV4 or IPV6.
+ * The third not void item must be TCP.
+ * The next not void item must be END.
+ * action:
+ * The first not void action should be QUEUE.
+ * The next not void action should be END.
+ * pattern example:
+ * ITEMSpecMask
+ * ETH NULLNULL
+ * IPV4/IPV6   NULLNULL
+ * TCP tcp_flags   0x020xFF
+ * END
+ * other members in mask and spec should set to 0x00.
+ * item->last should be NULL.
+ */
+static int
+cons_parse_syn_filter(const struct rte_flow_attr *attr,
+   const struct rte_flow_item pattern[],
+   const struct rte_flow_action actions[],
+   struct rte_eth_syn_filter *filter,
+   struct rte_flow_error *error)
+{
+   const struct rte_flow_item *item;
+   const struct rte_flow_action *act;
+   const struct rte_flow_item_tcp *tcp_spec;
+   const struct rte_flow_item_tcp *tcp_mask;
+   const struct rte_flow_action_queue *act_q;
+   uint32_t index;
+
+   if (!pattern) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM_NUM,
+   NULL, "NULL pattern.");
+   return -rte_errno;
+   }
+
+   if (!actions) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ACTION_NUM,
+   NULL, "NULL action.");
+   return -rte_errno;
+   }
+
+   if (!attr) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ATTR,
+  NULL, "NULL attribute.");
+   return -rte_errno;
+   }
+
+   /* parse pattern */
+   index = 0;
+
+   /* the first not void item should be MAC or IPv4 or IPv6 or TCP */
+   NEXT_ITEM_OF_PATTERN(item, pattern, index);
+   if (item->type != RTE_FLOW_ITEM_TYPE_ETH &&
+   item->type != RTE_FLOW_ITEM_TYPE_IPV4 &&
+   item->type != RTE_FLOW_ITEM_TYPE_IPV6 &&
+   item->type != RTE_FLOW_ITEM_TYPE_TCP) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM,
+   item, "Not supported by syn filter");
+   return -rte_errno;
+   }
+   /*Not supported last point for range*/
+   if (item->last) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+   item, "Not supported last point for range");
+   return -rte_errno;
+   }
+
+   /* Skip Ethernet */
+   if (item->type == RTE_FLOW_ITEM_TYPE_ETH) {
+   /* if the item is MAC, the content should be NULL */
+   if (item->spec || item->mask) {
+   rte_flow_error_set(error, EINVAL,
+   RTE_FLOW_ERROR_TYPE_ITEM,
+   item, "Invalid SYN address mask");
+   return -rte_errno;
+   }
+
+   /* check if the next not void item is IPv4 or IPv6 */
+   index++;
+  

[dpdk-dev] [PATCH v6 15/18] net/ixgbe: parse flow director filter

[Wei Zhao]

check if the rule is a flow director rule, and get the flow director info.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |2 +
 drivers/net/ixgbe/ixgbe_ethdev.h |   16 +
 drivers/net/ixgbe/ixgbe_fdir.c   |  253 +---
 drivers/net/ixgbe/ixgbe_flow.c   | 1252 +-
 4 files changed, 1414 insertions(+), 109 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 14a88ee..4b9f4aa 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -1436,6 +1436,8 @@ static int ixgbe_fdir_filter_init(struct rte_eth_dev 
*eth_dev)
 "Failed to allocate memory for fdir hash map!");
return -ENOMEM;
}
+   fdir_info->mask_added = FALSE;
+
return 0;
 }
 
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index 5efa650..dee159f 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -162,6 +162,17 @@ struct ixgbe_fdir_filter {
 /* list of fdir filters */
 TAILQ_HEAD(ixgbe_fdir_filter_list, ixgbe_fdir_filter);
 
+struct ixgbe_fdir_rule {
+   struct ixgbe_hw_fdir_mask mask;
+   union ixgbe_atr_input ixgbe_fdir; /* key of fdir filter*/
+   bool b_spec; /* If TRUE, ixgbe_fdir, fdirflags, queue have meaning. */
+   bool b_mask; /* If TRUE, mask has meaning. */
+   enum rte_fdir_mode mode; /* IP, MAC VLAN, Tunnel */
+   uint32_t fdirflags; /* drop or forward */
+   uint32_t soft_id; /* an unique value for this rule */
+   uint8_t queue; /* assigned rx queue */
+};
+
 struct ixgbe_hw_fdir_info {
struct ixgbe_hw_fdir_mask mask;
uint8_t flex_bytes_offset;
@@ -177,6 +188,7 @@ struct ixgbe_hw_fdir_info {
/* store the pointers of the filters, index is the hash value. */
struct ixgbe_fdir_filter **hash_map;
struct rte_hash *hash_handle; /* cuckoo hash handler */
+   bool mask_added; /* If already got mask from consistent filter */
 };
 
 /* structure for interrupt relative data */
@@ -479,6 +491,10 @@ bool ixgbe_rss_update_sp(enum ixgbe_mac_type mac_type);
  * Flow director function prototypes
  */
 int ixgbe_fdir_configure(struct rte_eth_dev *dev);
+int ixgbe_fdir_set_input_mask(struct rte_eth_dev *dev);
+int ixgbe_fdir_filter_program(struct rte_eth_dev *dev,
+ struct ixgbe_fdir_rule *rule,
+ bool del, bool update);
 
 void ixgbe_configure_dcb(struct rte_eth_dev *dev);
 
diff --git a/drivers/net/ixgbe/ixgbe_fdir.c b/drivers/net/ixgbe/ixgbe_fdir.c
index e928ad7..3b9d60c 100644
--- a/drivers/net/ixgbe/ixgbe_fdir.c
+++ b/drivers/net/ixgbe/ixgbe_fdir.c
@@ -112,10 +112,8 @@
 static int fdir_erase_filter_82599(struct ixgbe_hw *hw, uint32_t fdirhash);
 static int fdir_set_input_mask(struct rte_eth_dev *dev,
   const struct rte_eth_fdir_masks *input_mask);
-static int fdir_set_input_mask_82599(struct rte_eth_dev *dev,
-   const struct rte_eth_fdir_masks *input_mask);
-static int fdir_set_input_mask_x550(struct rte_eth_dev *dev,
-   const struct rte_eth_fdir_masks 
*input_mask);
+static int fdir_set_input_mask_82599(struct rte_eth_dev *dev);
+static int fdir_set_input_mask_x550(struct rte_eth_dev *dev);
 static int ixgbe_set_fdir_flex_conf(struct rte_eth_dev *dev,
const struct rte_eth_fdir_flex_conf *conf, uint32_t *fdirctrl);
 static int fdir_enable_82599(struct ixgbe_hw *hw, uint32_t fdirctrl);
@@ -295,8 +293,7 @@ reverse_fdir_bitmasks(uint16_t hi_dword, uint16_t lo_dword)
  * but makes use of the rte_fdir_masks structure to see which bits to set.
  */
 static int
-fdir_set_input_mask_82599(struct rte_eth_dev *dev,
-   const struct rte_eth_fdir_masks *input_mask)
+fdir_set_input_mask_82599(struct rte_eth_dev *dev)
 {
struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
struct ixgbe_hw_fdir_info *info =
@@ -308,8 +305,6 @@ fdir_set_input_mask_82599(struct rte_eth_dev *dev,
uint32_t fdirm = IXGBE_FDIRM_POOL | IXGBE_FDIRM_DIPv6 | 
IXGBE_FDIRM_FLEX;
uint32_t fdirtcpm;  /* TCP source and destination port masks. */
uint32_t fdiripv6m; /* IPv6 source and destination masks. */
-   uint16_t dst_ipv6m = 0;
-   uint16_t src_ipv6m = 0;
volatile uint32_t *reg;
 
PMD_INIT_FUNC_TRACE();
@@ -320,31 +315,30 @@ fdir_set_input_mask_82599(struct rte_eth_dev *dev,
 * a VLAN of 0 is unspecified, so mask that out as well.  L4type
 * cannot be masked out in this implementation.
 */
-   if (input_mask->dst_port_mask == 0 && input_mask->src_port_mask == 0)
+   if (info->mask.dst_port_mask == 0 && info->mask.src_port_mask == 0)
/* use the L4 protocol mask for raw IPv4/IPv6 traffic */
fdirm |= IXGBE_FDIRM_L4P;
 
-   i

[dpdk-dev] [PATCH v6 16/18] net/ixgbe: create consistent filter

[Wei Zhao]

This patch adds a function to create the flow directory filter.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |  25 +++--
 drivers/net/ixgbe/ixgbe_ethdev.h |  61 
 drivers/net/ixgbe/ixgbe_flow.c   | 194 ++-
 3 files changed, 266 insertions(+), 14 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 4b9f4aa..3648e12 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -305,9 +305,6 @@ static void ixgbevf_add_mac_addr(struct rte_eth_dev *dev,
 static void ixgbevf_remove_mac_addr(struct rte_eth_dev *dev, uint32_t index);
 static void ixgbevf_set_default_mac_addr(struct rte_eth_dev *dev,
 struct ether_addr *mac_addr);
-static int ixgbe_syn_filter_set(struct rte_eth_dev *dev,
-   struct rte_eth_syn_filter *filter,
-   bool add);
 static int ixgbe_syn_filter_get(struct rte_eth_dev *dev,
struct rte_eth_syn_filter *filter);
 static int ixgbe_syn_filter_handle(struct rte_eth_dev *dev,
@@ -317,17 +314,11 @@ static int ixgbe_add_5tuple_filter(struct rte_eth_dev 
*dev,
struct ixgbe_5tuple_filter *filter);
 static void ixgbe_remove_5tuple_filter(struct rte_eth_dev *dev,
struct ixgbe_5tuple_filter *filter);
-static int ixgbe_add_del_ntuple_filter(struct rte_eth_dev *dev,
-   struct rte_eth_ntuple_filter *filter,
-   bool add);
 static int ixgbe_ntuple_filter_handle(struct rte_eth_dev *dev,
enum rte_filter_op filter_op,
void *arg);
 static int ixgbe_get_ntuple_filter(struct rte_eth_dev *dev,
struct rte_eth_ntuple_filter *filter);
-static int ixgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
-   struct rte_eth_ethertype_filter *filter,
-   bool add);
 static int ixgbe_ethertype_filter_handle(struct rte_eth_dev *dev,
enum rte_filter_op filter_op,
void *arg);
@@ -1292,6 +1283,14 @@ eth_ixgbe_dev_init(struct rte_eth_dev *eth_dev)
 
/* initialize l2 tunnel filter list & hash */
ixgbe_l2_tn_filter_init(eth_dev);
+
+   TAILQ_INIT(&filter_ntuple_list);
+   TAILQ_INIT(&filter_ethertype_list);
+   TAILQ_INIT(&filter_syn_list);
+   TAILQ_INIT(&filter_fdir_list);
+   TAILQ_INIT(&filter_l2_tunnel_list);
+   TAILQ_INIT(&ixgbe_flow_list);
+
return 0;
 }
 
@@ -5742,7 +5741,7 @@ ixgbevf_set_default_mac_addr(struct rte_eth_dev *dev, 
struct ether_addr *addr)
return -ENOTSUP;\
 } while (0)
 
-static int
+int
 ixgbe_syn_filter_set(struct rte_eth_dev *dev,
struct rte_eth_syn_filter *filter,
bool add)
@@ -6121,7 +6120,7 @@ ntuple_filter_to_5tuple(struct rte_eth_ntuple_filter 
*filter,
  *- On success, zero.
  *- On failure, a negative value.
  */
-static int
+int
 ixgbe_add_del_ntuple_filter(struct rte_eth_dev *dev,
struct rte_eth_ntuple_filter *ntuple_filter,
bool add)
@@ -6266,7 +6265,7 @@ ixgbe_ntuple_filter_handle(struct rte_eth_dev *dev,
return ret;
 }
 
-static int
+int
 ixgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
struct rte_eth_ethertype_filter *filter,
bool add)
@@ -7345,7 +7344,7 @@ ixgbe_remove_l2_tn_filter(struct ixgbe_l2_tn_info 
*l2_tn_info,
 }
 
 /* Add l2 tunnel filter */
-static int
+int
 ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
   struct rte_eth_l2_tunnel_conf *l2_tunnel,
   bool restore)
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index dee159f..6908c3d 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -329,6 +329,54 @@ struct ixgbe_l2_tn_info {
bool e_tag_ether_type; /* ether type for e-tag */
 };
 
+struct rte_flow {
+   enum rte_filter_type filter_type;
+   void *rule;
+};
+/* ntuple filter list structure */
+struct ixgbe_ntuple_filter_ele {
+   TAILQ_ENTRY(ixgbe_ntuple_filter_ele) entries;
+   struct rte_eth_ntuple_filter filter_info;
+};
+/* ethertype filter list structure */
+struct ixgbe_ethertype_filter_ele {
+   TAILQ_ENTRY(ixgbe_ethertype_filter_ele) entries;
+   struct rte_eth_ethertype_filter filter_info;
+};
+/* syn filter list structure */
+struct ixgbe_eth_syn_filter_ele {
+   TAILQ_ENTRY(ixgbe_eth_syn_filter_ele) entries;
+   struct rte_eth_syn_filter filter_info;
+};
+/* fdir filter list structure */
+struct ixgbe_fdir_rule_ele {
+   TAILQ_ENTRY(ixgbe_fdir_rule_ele) entries;
+   struct ixgbe_fdir_rule filter_info;
+};
+/* l2_tu

[dpdk-dev] [PATCH v6 18/18] net/ixgbe: flush all the filter list

[Wei Zhao]

This patch adds a function to flush all the fliter list
filter on a port.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |  3 +++
 drivers/net/ixgbe/ixgbe_ethdev.h |  1 +
 drivers/net/ixgbe/ixgbe_flow.c   | 56 
 3 files changed, 60 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 1112a3e..f46507b 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -1341,6 +1341,9 @@ eth_ixgbe_dev_uninit(struct rte_eth_dev *eth_dev)
/* Remove all ntuple filters of the device */
ixgbe_ntuple_filter_uninit(eth_dev);
 
+   /* clear all the filters list */
+   ixgbe_filterlist_flush();
+
return 0;
 }
 
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index 01c18cd..5b31149 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -551,6 +551,7 @@ ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
 int
 ixgbe_dev_l2_tunnel_filter_del(struct rte_eth_dev *dev,
   struct rte_eth_l2_tunnel_conf *l2_tunnel);
+void ixgbe_filterlist_flush(void);
 /*
  * Flow director function prototypes
  */
diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
index 20aefa6..82aceed 100644
--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c
@@ -2444,6 +2444,60 @@ ixgbe_parse_fdir_filter(const struct rte_flow_attr *attr,
return ret;
 }
 
+void
+ixgbe_filterlist_flush(void)
+{
+   struct ixgbe_ntuple_filter_ele *ntuple_filter_ptr;
+   struct ixgbe_ethertype_filter_ele *ethertype_filter_ptr;
+   struct ixgbe_eth_syn_filter_ele *syn_filter_ptr;
+   struct ixgbe_eth_l2_tunnel_conf_ele *l2_tn_filter_ptr;
+   struct ixgbe_fdir_rule_ele *fdir_rule_ptr;
+   struct ixgbe_flow_mem *ixgbe_flow_mem_ptr;
+
+   while ((ntuple_filter_ptr = TAILQ_FIRST(&filter_ntuple_list))) {
+   TAILQ_REMOVE(&filter_ntuple_list,
+ntuple_filter_ptr,
+entries);
+   rte_free(ntuple_filter_ptr);
+   }
+
+   while ((ethertype_filter_ptr = TAILQ_FIRST(&filter_ethertype_list))) {
+   TAILQ_REMOVE(&filter_ethertype_list,
+ethertype_filter_ptr,
+entries);
+   rte_free(ethertype_filter_ptr);
+   }
+
+   while ((syn_filter_ptr = TAILQ_FIRST(&filter_syn_list))) {
+   TAILQ_REMOVE(&filter_syn_list,
+syn_filter_ptr,
+entries);
+   rte_free(syn_filter_ptr);
+   }
+
+   while ((l2_tn_filter_ptr = TAILQ_FIRST(&filter_l2_tunnel_list))) {
+   TAILQ_REMOVE(&filter_l2_tunnel_list,
+l2_tn_filter_ptr,
+entries);
+   rte_free(l2_tn_filter_ptr);
+   }
+
+   while ((fdir_rule_ptr = TAILQ_FIRST(&filter_fdir_list))) {
+   TAILQ_REMOVE(&filter_fdir_list,
+fdir_rule_ptr,
+entries);
+   rte_free(fdir_rule_ptr);
+   }
+
+   while ((ixgbe_flow_mem_ptr = TAILQ_FIRST(&ixgbe_flow_list))) {
+   TAILQ_REMOVE(&ixgbe_flow_list,
+ixgbe_flow_mem_ptr,
+entries);
+   rte_free(ixgbe_flow_mem_ptr->flow);
+   rte_free(ixgbe_flow_mem_ptr);
+   }
+}
+
 /**
  * Create or destroy a flow rule.
  * Theorically one rule can match more than one filters.
@@ -2818,5 +2872,7 @@ ixgbe_flow_flush(struct rte_eth_dev *dev,
return ret;
}
 
+   ixgbe_filterlist_flush();
+
return 0;
 }
-- 
2.5.5

[dpdk-dev] [PATCH v6 17/18] net/ixgbe: destroy consistent filter

[Wei Zhao]

This patch adds a function to destroy the flow fliter.

Signed-off-by: Wei Zhao 
Signed-off-by: Wenzhuo Lu 
---
 drivers/net/ixgbe/ixgbe_ethdev.c |   2 +-
 drivers/net/ixgbe/ixgbe_ethdev.h |   3 +
 drivers/net/ixgbe/ixgbe_flow.c   | 117 ++-
 3 files changed, 120 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 3648e12..1112a3e 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -7401,7 +7401,7 @@ ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
 }
 
 /* Delete l2 tunnel filter */
-static int
+int
 ixgbe_dev_l2_tunnel_filter_del(struct rte_eth_dev *dev,
   struct rte_eth_l2_tunnel_conf *l2_tunnel)
 {
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
index 6908c3d..01c18cd 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.h
+++ b/drivers/net/ixgbe/ixgbe_ethdev.h
@@ -548,6 +548,9 @@ int
 ixgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
   struct rte_eth_l2_tunnel_conf *l2_tunnel,
   bool restore);
+int
+ixgbe_dev_l2_tunnel_filter_del(struct rte_eth_dev *dev,
+  struct rte_eth_l2_tunnel_conf *l2_tunnel);
 /*
  * Flow director function prototypes
  */
diff --git a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c
index 16b43d9..20aefa6 100644
--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c
@@ -162,11 +162,14 @@ static struct rte_flow *ixgbe_flow_create(struct 
rte_eth_dev *dev,
const struct rte_flow_item pattern[],
const struct rte_flow_action actions[],
struct rte_flow_error *error);
+static int ixgbe_flow_destroy(struct rte_eth_dev *dev,
+   struct rte_flow *flow,
+   struct rte_flow_error *error);
 
 const struct rte_flow_ops ixgbe_flow_ops = {
ixgbe_flow_validate,
ixgbe_flow_create,
-   NULL,
+   ixgbe_flow_destroy,
ixgbe_flow_flush,
NULL,
 };
@@ -2678,6 +2681,118 @@ ixgbe_flow_validate(__rte_unused struct rte_eth_dev 
*dev,
return ret;
 }
 
+/* Destroy a flow rule on ixgbe. */
+static int
+ixgbe_flow_destroy(struct rte_eth_dev *dev,
+   struct rte_flow *flow,
+   struct rte_flow_error *error)
+{
+   int ret;
+   struct rte_flow *pmd_flow = flow;
+   enum rte_filter_type filter_type = pmd_flow->filter_type;
+   struct rte_eth_ntuple_filter ntuple_filter;
+   struct rte_eth_ethertype_filter ethertype_filter;
+   struct rte_eth_syn_filter syn_filter;
+   struct ixgbe_fdir_rule fdir_rule;
+   struct rte_eth_l2_tunnel_conf l2_tn_filter;
+   struct ixgbe_ntuple_filter_ele *ntuple_filter_ptr;
+   struct ixgbe_ethertype_filter_ele *ethertype_filter_ptr;
+   struct ixgbe_eth_syn_filter_ele *syn_filter_ptr;
+   struct ixgbe_eth_l2_tunnel_conf_ele *l2_tn_filter_ptr;
+   struct ixgbe_fdir_rule_ele *fdir_rule_ptr;
+   struct ixgbe_flow_mem *ixgbe_flow_mem_ptr;
+
+   switch (filter_type) {
+   case RTE_ETH_FILTER_NTUPLE:
+   ntuple_filter_ptr = (struct ixgbe_ntuple_filter_ele *)
+   pmd_flow->rule;
+   (void)rte_memcpy(&ntuple_filter,
+   &ntuple_filter_ptr->filter_info,
+   sizeof(struct rte_eth_ntuple_filter));
+   ret = ixgbe_add_del_ntuple_filter(dev, &ntuple_filter, FALSE);
+   if (!ret) {
+   TAILQ_REMOVE(&filter_ntuple_list,
+   ntuple_filter_ptr, entries);
+   rte_free(ntuple_filter_ptr);
+   }
+   break;
+   case RTE_ETH_FILTER_ETHERTYPE:
+   ethertype_filter_ptr = (struct ixgbe_ethertype_filter_ele *)
+   pmd_flow->rule;
+   (void)rte_memcpy(ðertype_filter,
+   ðertype_filter_ptr->filter_info,
+   sizeof(struct rte_eth_ethertype_filter));
+   ret = ixgbe_add_del_ethertype_filter(dev,
+   ðertype_filter, FALSE);
+   if (!ret) {
+   TAILQ_REMOVE(&filter_ethertype_list,
+   ethertype_filter_ptr, entries);
+   rte_free(ethertype_filter_ptr);
+   }
+   break;
+   case RTE_ETH_FILTER_SYN:
+   syn_filter_ptr = (struct ixgbe_eth_syn_filter_ele *)
+   pmd_flow->rule;
+   (void)rte_memcpy(&syn_filter,
+   &syn_filter_ptr->filter_info,
+   sizeof(struct rte_eth_syn_filter));
+   ret = ixgbe_syn_filter_set(dev, &syn_filter, FALSE);
+   if (!ret) {
+   TAILQ_REMOVE(&filter_syn_list,
+ 

Re: [dpdk-dev] [PATCH v3 2/8] lib: add cryptodev type for the upcoming ARMv8 PMD

[Hemant Agrawal]

On 1/4/2017 11:03 PM, zbigniew.bo...@caviumnetworks.com wrote:

From: Zbigniew Bodek 

Add type and name for ARMv8 crypto PMD

Signed-off-by: Zbigniew Bodek 
---
 lib/librte_cryptodev/rte_cryptodev.h | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/lib/librte_cryptodev/rte_cryptodev.h 
b/lib/librte_cryptodev/rte_cryptodev.h
index 8f63e8f..6f34f22 100644
--- a/lib/librte_cryptodev/rte_cryptodev.h
+++ b/lib/librte_cryptodev/rte_cryptodev.h
@@ -66,6 +66,8 @@
 /**< KASUMI PMD device name */
 #define CRYPTODEV_NAME_ZUC_PMD crypto_zuc
 /**< KASUMI PMD device name */
+#define CRYPTODEV_NAME_ARMV8_PMD   crypto_armv8
+/**< ARMv8 Crypto PMD device name */


I will suggest the name as armv8ce or armv8_ce for this driver.
Do you agree?


 /** Crypto device type */
 enum rte_cryptodev_type {
@@ -77,6 +79,7 @@ enum rte_cryptodev_type {
RTE_CRYPTODEV_KASUMI_PMD,   /**< KASUMI PMD */
RTE_CRYPTODEV_ZUC_PMD,  /**< ZUC PMD */
RTE_CRYPTODEV_OPENSSL_PMD,/**<  OpenSSL PMD */
+   RTE_CRYPTODEV_ARMV8_PMD,/**< ARMv8 crypto PMD */
 };

 extern const char **rte_cyptodev_names;

Re: [dpdk-dev] [PATCH v3 16/29] net/bnxt: use eal I/O device memory read/write API

[Jerin Jacob]

On Thu, Jan 12, 2017 at 07:10:24PM +, Ferruh Yigit wrote:

Thanks Ferruh for the review.


> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> <...>
> >  #define B_CP_DB_REARM(cpr, raw_cons)   
> > \
> > -   (*(uint32_t *)((cpr)->cp_doorbell) = (DB_CP_REARM_FLAGS | \
> > -   RING_CMP(cpr->cp_ring_struct, raw_cons)))
> > +   rte_write32((DB_CP_REARM_FLAGS |\
> 
> Just asking, can this be rte_write32_relaxed() since there is explicit
> memory barrier defined for B_CP_DIS_DB but not for here?

Here is the logic followed across the driver change:

1) If the top level code is in the pattern like "explicit barrier" and then
io write. Then second the io write operation can be relaxed.
2) If the code runs in slow path, To make patch clean and scope limited,
use only nonrelaxed versions.

> 
> > +   RING_CMP(((cpr)->cp_ring_struct), raw_cons)),   \
> > +   ((cpr)->cp_doorbell))
> >  
> >  #define B_CP_DIS_DB(cpr, raw_cons) \
> > -   rte_smp_wmb();  \
> > -   (*(uint32_t *)((cpr)->cp_doorbell) = (DB_CP_FLAGS | \
> > -   RING_CMP(cpr->cp_ring_struct, raw_cons)))
> > +   rte_write32((DB_CP_FLAGS |  \
> > +   RING_CMP(((cpr)->cp_ring_struct), raw_cons)),   \
> > +   ((cpr)->cp_doorbell))
> >  
> 
> <...>
> 
> > @@ -80,11 +82,12 @@ static int bnxt_hwrm_send_message_locked(struct bnxt 
> > *bp, void *msg,
> > for (; i < bp->max_req_len; i += 4) {
> > bar = (uint8_t *)bp->bar0 + i;
> > *(volatile uint32_t *)bar = 0;
> 
> Should this line be removed?

Yes. Will be fixed in next revision.

> 
> > +   rte_write32(0, bar);
> > }
> >  
> > /* Ring channel doorbell */
> > bar = (uint8_t *)bp->bar0 + 0x100;
> > -   *(volatile uint32_t *)bar = 1;
> > +   rte_write32(1, bar);
> >  
> > /* Poll for the valid bit */
> > for (i = 0; i < HWRM_CMD_TIMEOUT; i++) {
> <...>

Re: [dpdk-dev] [PATCH v3 17/29] net/bnx2x: use eal I/O device memory read/write API

[Jerin Jacob]

On Thu, Jan 12, 2017 at 07:11:18PM +, Ferruh Yigit wrote:
> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> <...>
> >  
> > @@ -1560,11 +1556,9 @@ bnx2x_reg_read32(struct bnx2x_softc *sc, size_t 
> > offset)
> >  #define DPM_TRIGGER_TYPE 0x40
> >  
> >  /* Doorbell macro */
> > -#define BNX2X_DB_WRITE(db_bar, val) \
> > -   *((volatile uint32_t *)(db_bar)) = (val)
> > +#define BNX2X_DB_WRITE(db_bar, val) rte_write32_relaxed((val), (db_bar))
> 
> What is the rule to use relaxed version or not?

Here is the logic followed across the driver change:

1) If the top level code is in the pattern like "explicit barrier" and
then io write. Then second the io write operation can be relaxed.
2) If the code runs in slow path, To make patch clean and scope limited,
use only nonrelaxed versions.
 
> I don't know about memory barrier requirements, if it is easy, would you
> mind explaining? Because I have same question for many different parts
> of this patchset.
> 
> Thanks,
> ferruh

Re: [dpdk-dev] [PATCH v3 18/29] net/cxgbe: use eal I/O device memory read/write API

[Jerin Jacob]

On Thu, Jan 12, 2017 at 07:12:33PM +, Ferruh Yigit wrote:
> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> <...>
> 
> >  
> > -#define CXGBE_PCI_REG_WRITE(reg, value) ({ \
> > -   CXGBE_PCI_REG((reg)) = (value); })
> > +#define CXGBE_PCI_REG_WRITE(reg, value) rte_write32((value), (reg))
> 
> Almost all (if not all) PMD write macros' argument order is like
> write(address, value), but rte_writeX has rte_writex(value, address)
> 
> What is the common usage for this kind of function?

Arguments order has been taken from Linux kernel readl/writel syntax.

> What do you think reverting argument order?

OMG :-). If it worth it then we can consider. IMHO, let it be in kernel
syntax so that it will easy to port drivers from Linux kernel.

> 
> As a similar example, dpdk ether_addr_copy(src, dst) function argument
> order is revers according common usage, and keeps confusing people.
> 
> <...>
> 

Re: [dpdk-dev] [PATCH v5] app/testpmd: supported offload capabilities query

[De Lara Guarch, Pablo]

Hi Qiming,

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Qiming Yang
> Sent: Thursday, January 12, 2017 3:26 AM
> To: dev@dpdk.org
> Cc: Yang, Qiming
> Subject: [dpdk-dev] [PATCH v5] app/testpmd: supported offload capabilities
> query
> 
> Add two new commands "show port cap " and "show
> port cap all"to diaplay what offload capabilities supported
> in ports. It will not only display all the capabilities of
> the port, but also the enabling condition for each capability
> in the running time.
> 
> Signed-off-by: Qiming Yang 
> Acked-by: Jingjing Wu 
> Acked-by: Beilei Xing 
> ---
> v2 changes:
> * fixed the output style as Ferruh's patch show and add some
>   description in docs for new functions.
> v3 changes:
> * add new command in cmd_help_long_parsed.
> v4 changes:
> * use 'cap' instead of 'capa'.
> v5 changes:
> * rebased, fixed the inappropriate expression and adjusted the
>   output order.
> ---
> ---
>  app/test-pmd/cmdline.c  |  17 ++-
>  app/test-pmd/config.c   | 175
> 
>  app/test-pmd/testpmd.h  |   1 +
>  doc/guides/testpmd_app_ug/testpmd_funcs.rst |  12 +-
>  4 files changed, 195 insertions(+), 10 deletions(-)
> 
> diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
> index 4e8b0d8..6fa1783 100644
> --- a/app/test-pmd/cmdline.c
> +++ b/app/test-pmd/cmdline.c

...

>  cmdline_parse_token_string_t cmd_showportall_show =
> @@ -5821,13 +5824,14 @@ cmdline_parse_token_string_t
> cmd_showportall_port =
>   TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, port,
> "port");
>  cmdline_parse_token_string_t cmd_showportall_what =
>   TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, what,
> -
> "info#stats#xstats#fdir#stat_qmap#dcb_tc");
> +
> "info#stats#xstats#fdir#stat_qmap#dcb_tc#cap");
>  cmdline_parse_token_string_t cmd_showportall_all =
>   TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, all,
> "all");
>  cmdline_parse_inst_t cmd_showportall = {
>   .f = cmd_showportall_parsed,
>   .data = NULL,
> - .help_str = "show|clear port
> info|stats|xstats|fdir|stat_qmap|dcb_tc all",
> + .help_str = "show|clear port"

Missing space after "port"

> + "info|stats|xstats|fdir|stat_qmap|dcb_tc|cap all",
>   .tokens = {
>   (void *)&cmd_showportall_show,
>   (void *)&cmd_showportall_port,
> @@ -5867,6 +5871,8 @@ static void cmd_showport_parsed(void
> *parsed_result,
>   nic_stats_mapping_display(res->portnum);
>   else if (!strcmp(res->what, "dcb_tc"))
>   port_dcb_info_display(res->portnum);
> + else if (!strcmp(res->what, "cap"))
> + port_offload_cap_display(res->portnum);
>  }
> 
>  cmdline_parse_token_string_t cmd_showport_show =
> @@ -5876,14 +5882,15 @@ cmdline_parse_token_string_t
> cmd_showport_port =
>   TOKEN_STRING_INITIALIZER(struct cmd_showport_result, port,
> "port");
>  cmdline_parse_token_string_t cmd_showport_what =
>   TOKEN_STRING_INITIALIZER(struct cmd_showport_result, what,
> -
> "info#stats#xstats#fdir#stat_qmap#dcb_tc");
> +
> "info#stats#xstats#fdir#stat_qmap#dcb_tc#cap");
>  cmdline_parse_token_num_t cmd_showport_portnum =
>   TOKEN_NUM_INITIALIZER(struct cmd_showport_result, portnum,
> UINT8);
> 
>  cmdline_parse_inst_t cmd_showport = {
>   .f = cmd_showport_parsed,
>   .data = NULL,
> - .help_str = "show|clear port
> info|stats|xstats|fdir|stat_qmap|dcb_tc "
> + .help_str = "show|clear port"

Missing space after "port".
Also, add a tab in the following line to be consistent with
the help_str of the next command.

> + "info|stats|xstats|fdir|stat_qmap|dcb_tc|cap "
>   "",
>   .tokens = {
>   (void *)&cmd_showport_show,
> diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c
> index 617e6d4..8a2ef56 100644
> --- a/app/test-pmd/config.c
> +++ b/app/test-pmd/config.c

...

> +
> + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_CKSUM) {
> + printf("RX TCP checksum:   ");
> + if (dev->data->dev_conf.rxmode.hw_ip_checksum)
> + printf("on\n");
> + else
> + printf("off\n");
> + }
> +
> + if (dev_info.rx_offload_capa &
> DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM)
> + printf("RX Outer IPv4 checksum:");

Missing on/off?

> +
> + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_LRO) {
> + printf("Large receive offload: ");
> + if (dev->data->dev_conf.rxmode.enable_lro)
> + printf("on\n");
> + else
> + printf("off\n");
> + }

Thanks for this nice patch.

Pablo

Re: [dpdk-dev] [PATCH] examples/ipsec-secgw: add support for SHA256 HMAC

[Sergio Gonzalez Monroy]

On 12/01/2017 14:52, zbigniew.bo...@caviumnetworks.com wrote:

From: Zbigniew Bodek 

Add minor adjustments to support SHA256 HMAC:
- extend maximum key length to match SHA256 HMAC
- add SHA256 HMAC parameters and configuration string
- add SHA256 HMAC to inbound and outbound cases

Signed-off-by: Zbigniew Bodek 
---


Acked-by: Sergio Gonzalez Monroy 

Re: [dpdk-dev] [PATCH v6 5/6] app/testpmd: add MACsec offload commands

[Thomas Monjalon]

There is an error with clang:

app/test-pmd/cmdline.c:11535:13: fatal error:
expression which evaluates to zero treated as a null pointer
constant of type 'const char *' [-Wnon-literal-null-conversion]
 port_id, UINT8);
  ^

Please, send a v7

Re: [dpdk-dev] [PATCH v6 5/6] app/testpmd: add MACsec offload commands

[Tiwei Bie]

On Fri, Jan 13, 2017 at 10:07:34AM +0100, Thomas Monjalon wrote:
> There is an error with clang:
> 
> app/test-pmd/cmdline.c:11535:13: fatal error:
> expression which evaluates to zero treated as a null pointer
> constant of type 'const char *' [-Wnon-literal-null-conversion]
>  port_id, UINT8);
>   ^
> 
> Please, send a v7

Oops.. Really sorry about it.. Will send v7!

Best regards,
Tiwei Bie

Re: [dpdk-dev] [PATCH v3 8/8] app/test: add ARMv8 crypto tests and test vectors

[Hemant Agrawal]

On 1/4/2017 11:03 PM, zbigniew.bo...@caviumnetworks.com wrote:

From: Zbigniew Bodek 

Introduce unit tests for ARMv8 crypto PMD.
Add test vectors for short cases such as 160 bytes.
These test cases are ARMv8 specific since the code provides
different processing paths for different input data sizes.

User can validate correctness of algorithms' implementation using:
* cryptodev_sw_armv8_autotest
For performance test one can use:
* cryptodev_sw_armv8_perftest

Signed-off-by: Zbigniew Bodek 
---
 app/test/test_cryptodev.c  |  63 
 app/test/test_cryptodev_aes_test_vectors.h | 144 -
 app/test/test_cryptodev_blockcipher.c  |   4 +
 app/test/test_cryptodev_blockcipher.h  |   1 +
 app/test/test_cryptodev_perf.c | 480 +
 5 files changed, 684 insertions(+), 8 deletions(-)

diff --git a/app/test/test_cryptodev.c b/app/test/test_cryptodev.c
index 872f8b4..a0540d6 100644
--- a/app/test/test_cryptodev.c
+++ b/app/test/test_cryptodev.c



@@ -2422,6 +2449,136 @@ struct crypto_data_params 
aes_cbc_hmac_sha256_output[MAX_PACKET_SIZE_INDEX] = {
return TEST_SUCCESS;
 }

+static int
+test_perf_armv8_optimise_cyclecount(struct perf_test_params *pparams)
+{
+   uint32_t num_to_submit = pparams->total_operations;
+   struct rte_crypto_op *c_ops[num_to_submit];
+   struct rte_crypto_op *proc_ops[num_to_submit];
+   uint64_t failed_polls, retries, start_cycles, end_cycles,
+total_cycles = 0;
+   uint32_t burst_sent = 0, burst_received = 0;
+   uint32_t i, burst_size, num_sent, num_ops_received;
+
+   struct crypto_testsuite_params *ts_params = &testsuite_params;
+
+   static struct rte_cryptodev_sym_session *sess;
+
+   unsigned int digest_length = get_auth_digest_length(pparams->auth_algo);
+
+   if (rte_cryptodev_count() == 0) {
+   printf("\nNo crypto devices found. Is PMD build configured?\n");
+   return TEST_FAILED;
+   }
+
+   /* Create Crypto session*/
+   sess = test_perf_create_armv8_session(ts_params->dev_id,
+   pparams->chain, pparams->cipher_algo,
+   pparams->cipher_key_length, pparams->auth_algo);
+   TEST_ASSERT_NOT_NULL(sess, "Session creation failed");
+
+   /* Generate Crypto op data structure(s)*/
+   for (i = 0; i < num_to_submit ; i++) {
+   struct rte_mbuf *m = test_perf_create_pktmbuf(
+   ts_params->mbuf_mp,
+   pparams->buf_size);
+   TEST_ASSERT_NOT_NULL(m, "Failed to allocate tx_buf");
+
+   struct rte_crypto_op *op =
+   rte_crypto_op_alloc(ts_params->op_mpool,
+   RTE_CRYPTO_OP_TYPE_SYMMETRIC);
+   TEST_ASSERT_NOT_NULL(op, "Failed to allocate op");
+
+   op = test_perf_set_crypto_op_aes(op, m, sess, pparams->buf_size,
+   digest_length);
+   TEST_ASSERT_NOT_NULL(op, "Failed to attach op to session");
+
+   c_ops[i] = op;
+   }
+
+   printf("\nOn %s dev%u qp%u, %s, cipher algo:%s, cipher key length:%u, "
+   "auth_algo:%s, Packet Size %u bytes",
+   pmd_name(gbl_cryptodev_perftest_devtype),
+   ts_params->dev_id, 0,
+   chain_mode_name(pparams->chain),
+   cipher_algo_name(pparams->cipher_algo),
+   pparams->cipher_key_length,
+   auth_algo_name(pparams->auth_algo),
+   pparams->buf_size);
+   printf("\nOps Tx\tOps Rx\tOps/burst  ");
+   printf("Retries  "
+   "EmptyPolls\tIACycles/CyOp\tIACycles/Burst\tIACycles/Byte");
+
+   for (i = 2; i <= 128 ; i *= 2) {
+   num_sent = 0;
+   num_ops_received = 0;
+   retries = 0;
+   failed_polls = 0;
+   burst_size = i;
+   total_cycles = 0;
+   while (num_sent < num_to_submit) {
+   start_cycles = rte_rdtsc_precise();
+   burst_sent = rte_cryptodev_enqueue_burst(
+   ts_params->dev_id,
+   0, &c_ops[num_sent],
+   ((num_to_submit - num_sent) < burst_size) ?
+   num_to_submit - num_sent : burst_size);
+   end_cycles = rte_rdtsc_precise();
+   if (burst_sent == 0)
+   retries++;
+   num_sent += burst_sent;
+   total_cycles += (end_cycles - start_cycles);
+
+   /* Wait until requests have been sent. */
+   rte_delay_ms(1);
+

you may remove this delay.


+   start_cycles = rt

Re: [dpdk-dev] [PATCH v5] app/testpmd: supported offload capabilities query

[Yang, Qiming]

Hi, Pablo

-Original Message-
From: De Lara Guarch, Pablo 
Sent: Friday, January 13, 2017 4:43 PM
To: Yang, Qiming ; dev@dpdk.org
Cc: Yang, Qiming 
Subject: RE: [dpdk-dev] [PATCH v5] app/testpmd: supported offload capabilities 
query

Hi Qiming,

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Qiming Yang
> Sent: Thursday, January 12, 2017 3:26 AM
> To: dev@dpdk.org
> Cc: Yang, Qiming
> Subject: [dpdk-dev] [PATCH v5] app/testpmd: supported offload 
> capabilities query
> 
> Add two new commands "show port cap " and "show port cap all"to 
> diaplay what offload capabilities supported in ports. It will not only 
> display all the capabilities of the port, but also the enabling 
> condition for each capability in the running time.
> 
> Signed-off-by: Qiming Yang 
> Acked-by: Jingjing Wu 
> Acked-by: Beilei Xing 
> ---
> v2 changes:
> * fixed the output style as Ferruh's patch show and add some
>   description in docs for new functions.
> v3 changes:
> * add new command in cmd_help_long_parsed.
> v4 changes:
> * use 'cap' instead of 'capa'.
> v5 changes:
> * rebased, fixed the inappropriate expression and adjusted the
>   output order.
> ---
> ---
>  app/test-pmd/cmdline.c  |  17 ++-
>  app/test-pmd/config.c   | 175
> 
>  app/test-pmd/testpmd.h  |   1 +
>  doc/guides/testpmd_app_ug/testpmd_funcs.rst |  12 +-
>  4 files changed, 195 insertions(+), 10 deletions(-)
> 
> diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c index 
> 4e8b0d8..6fa1783 100644
> --- a/app/test-pmd/cmdline.c
> +++ b/app/test-pmd/cmdline.c

...

>  cmdline_parse_token_string_t cmd_showportall_show = @@ -5821,13 
> +5824,14 @@ cmdline_parse_token_string_t cmd_showportall_port =
>   TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, port, 
> "port");  cmdline_parse_token_string_t cmd_showportall_what =
>   TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, what,
> -
> "info#stats#xstats#fdir#stat_qmap#dcb_tc");
> +
> "info#stats#xstats#fdir#stat_qmap#dcb_tc#cap");
>  cmdline_parse_token_string_t cmd_showportall_all =
>   TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, all, "all");  
> cmdline_parse_inst_t cmd_showportall = {
>   .f = cmd_showportall_parsed,
>   .data = NULL,
> - .help_str = "show|clear port
> info|stats|xstats|fdir|stat_qmap|dcb_tc all",
> + .help_str = "show|clear port"

Missing space after "port"

> + "info|stats|xstats|fdir|stat_qmap|dcb_tc|cap all",
>   .tokens = {
>   (void *)&cmd_showportall_show,
>   (void *)&cmd_showportall_port,
> @@ -5867,6 +5871,8 @@ static void cmd_showport_parsed(void 
> *parsed_result,
>   nic_stats_mapping_display(res->portnum);
>   else if (!strcmp(res->what, "dcb_tc"))
>   port_dcb_info_display(res->portnum);
> + else if (!strcmp(res->what, "cap"))
> + port_offload_cap_display(res->portnum);
>  }
> 
>  cmdline_parse_token_string_t cmd_showport_show = @@ -5876,14 +5882,15 
> @@ cmdline_parse_token_string_t cmd_showport_port =
>   TOKEN_STRING_INITIALIZER(struct cmd_showport_result, port, "port");  
> cmdline_parse_token_string_t cmd_showport_what =
>   TOKEN_STRING_INITIALIZER(struct cmd_showport_result, what,
> -
> "info#stats#xstats#fdir#stat_qmap#dcb_tc");
> +
> "info#stats#xstats#fdir#stat_qmap#dcb_tc#cap");
>  cmdline_parse_token_num_t cmd_showport_portnum =
>   TOKEN_NUM_INITIALIZER(struct cmd_showport_result, portnum, UINT8);
> 
>  cmdline_parse_inst_t cmd_showport = {
>   .f = cmd_showport_parsed,
>   .data = NULL,
> - .help_str = "show|clear port
> info|stats|xstats|fdir|stat_qmap|dcb_tc "
> + .help_str = "show|clear port"

Missing space after "port".
Also, add a tab in the following line to be consistent with the help_str of the 
next command.

> + "info|stats|xstats|fdir|stat_qmap|dcb_tc|cap "
>   "",
>   .tokens = {
>   (void *)&cmd_showport_show,
> diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c index 
> 617e6d4..8a2ef56 100644
> --- a/app/test-pmd/config.c
> +++ b/app/test-pmd/config.c

...

> +
> + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_CKSUM) {
> + printf("RX TCP checksum:   ");
> + if (dev->data->dev_conf.rxmode.hw_ip_checksum)
> + printf("on\n");
> + else
> + printf("off\n");
> + }
> +
> + if (dev_info.rx_offload_capa &
> DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM)
> + printf("RX Outer IPv4 checksum:");

Missing on/off?
Qiming: I didn't find any switch for this feature in DPDK now. I'll fix the 
format problem. Thank you.
> +
> + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_LRO) {
> + printf("Large receive offload: ");
> + if (dev->data->dev_conf.rxmode.enable_lro)
> +

Re: [dpdk-dev] [PATCH v6 4/4] latencystats: added new library for latency stats

[Mcnamara, John]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Thomas Monjalon
> Sent: Thursday, January 12, 2017 1:41 PM
> To: Horton, Remy ; Pattan, Reshma
> 
> Cc: dev@dpdk.org; olivier.m...@6wind.com
> Subject: Re: [dpdk-dev] [PATCH v6 4/4] latencystats: added new library for
> latency stats
> 
> 2017-01-12 00:03, Remy Horton:
> > Add a library designed to calculate latency statistics and report them
> > to the application when queried. The library measures minimum, average
> > and maximum latencies, and jitter in nano seconds. The current
> > implementation supports global latency stats, i.e. per application
> stats.
> 
> Is it specific to ethdev ports?
> 
> > Added new field to mbuf struct to mark the packet arrival time on Rx.
> 
> There was another patch adding a timestamp in mbuf:
>   http://dpdk.org/ml/archives/dev/2016-October/048809.html

Hi Thomas,

Is that an issue? The other patch adds the same field for more or less the same 
purpose.
If the other patch goes in we can rework this patch or if this patch goes in the
other one won't be required.

John

Re: [dpdk-dev] [PATCH v6 4/4] latencystats: added new library for latency stats

[Thomas Monjalon]

2017-01-13 09:45, Mcnamara, John:
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Thomas Monjalon
> > 2017-01-12 00:03, Remy Horton:
> > > Add a library designed to calculate latency statistics and report them
> > > to the application when queried. The library measures minimum, average
> > > and maximum latencies, and jitter in nano seconds. The current
> > > implementation supports global latency stats, i.e. per application
> > stats.
> > 
> > Is it specific to ethdev ports?
> > 
> > > Added new field to mbuf struct to mark the packet arrival time on Rx.
> > 
> > There was another patch adding a timestamp in mbuf:
> > http://dpdk.org/ml/archives/dev/2016-October/048809.html
> 
> Hi Thomas,
> 
> Is that an issue? The other patch adds the same field for more or less the 
> same purpose.
> If the other patch goes in we can rework this patch or if this patch goes in 
> the
> other one won't be required.

Yes
The only issue is that it is not yet accepted or merged.
I think Olivier was considering it in a more global mbuf rework.

[dpdk-dev] [PATCH v2 0/4] enhancement to i40e PF host driver

[Chen Jing D(Mark)]

v2:
- add macro to replace numeric
- rework comments

Current PF host driver can serve DPDK VF well, but the
implementation is not complete to support Linux VF,
even both DPDK VF and Linux VF use same API set.

Note that the patch are experimental for use and might
be removed without prior notice.

This patch set made below changes:
1. Make an enhancement on interface to serve VF, so
   both Linux and DPDK VF can be well served.
2. Change API version number so both DPDK and Linux
   VF can recognize and select proper command and
   data structure to request service. But the
   sacrifice is DPDK VF can't identify host driver
   (Linux or DPDK) and extended function provided
   in DPDK PF host driver can't be used.
   This situation will change after negotiate with
   Linux maintainer to provide a better mechanism
   to identify both PF and VF function.

Chen Jing D(Mark) (4):
  net/i40e: change version number to support Linux VF
  net/i40e: return correct VSI id
  net/i40e: parse more VF parameter and configure
  net/i40e: support Linux VF to configure IRQ link list

 drivers/net/i40e/i40e_pf.c |  171 +++-
 1 files changed, 153 insertions(+), 18 deletions(-)

-- 
1.7.7.6

[dpdk-dev] [PATCH v2 2/4] net/i40e: return correct VSI id

[Chen Jing D(Mark)]

PF host didn't return correct VSI id to VF.
This change fix it.

Signed-off-by: Chen Jing D(Mark) 
---
 drivers/net/i40e/i40e_pf.c |3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/drivers/net/i40e/i40e_pf.c b/drivers/net/i40e/i40e_pf.c
index fa4af2b..fd83c16 100644
--- a/drivers/net/i40e/i40e_pf.c
+++ b/drivers/net/i40e/i40e_pf.c
@@ -329,8 +329,7 @@
 
/* Change below setting if PF host can support more VSIs for VF */
vf_res->vsi_res[0].vsi_type = I40E_VSI_SRIOV;
-   /* As assume Vf only has single VSI now, always return 0 */
-   vf_res->vsi_res[0].vsi_id = 0;
+   vf_res->vsi_res[0].vsi_id = vf->vsi->vsi_id;
vf_res->vsi_res[0].num_queue_pairs = vf->vsi->nb_qps;
ether_addr_copy(&vf->mac_addr,
(struct ether_addr *)vf_res->vsi_res[0].default_mac_addr);
-- 
1.7.7.6

[dpdk-dev] [PATCH v2 1/4] net/i40e: change version number to support Linux VF

[Chen Jing D(Mark)]

i40e PF host only support to work with DPDK VF driver, Linux
VF driver is not supported. This change will enhance in version
number returned.

Current version info returned won't be able to be recognized
by Linux VF driver, change to values that both DPDK VF and Linux
driver can recognize.

The expense is original DPDK host specific feature like
CFG_VLAN_PVID and CONFIG_VSI_QUEUES_EXT will not available.

DPDK VF also can't identify host driver by version number returned.
It always assume talking with Linux PF.

Signed-off-by: Chen Jing D(Mark) 
---
 drivers/net/i40e/i40e_pf.c |   12 ++--
 1 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/drivers/net/i40e/i40e_pf.c b/drivers/net/i40e/i40e_pf.c
index ddfc140..fa4af2b 100644
--- a/drivers/net/i40e/i40e_pf.c
+++ b/drivers/net/i40e/i40e_pf.c
@@ -276,8 +276,16 @@
 {
struct i40e_virtchnl_version_info info;
 
-   info.major = I40E_DPDK_VERSION_MAJOR;
-   info.minor = I40E_DPDK_VERSION_MINOR;
+   /* Respond like a Linux PF host in order to support both DPDK VF and
+* Linux VF driver. The expense is original DPDK host specific feature
+* like CFG_VLAN_PVID and CONFIG_VSI_QUEUES_EXT will not available.
+*
+* DPDK VF also can't identify host driver by version number returned.
+* It always assume talking with Linux PF.
+*/
+   info.major = I40E_VIRTCHNL_VERSION_MAJOR;
+   info.minor = I40E_VIRTCHNL_VERSION_MINOR_NO_VF_CAPS;
+
i40e_pf_host_send_msg_to_vf(vf, I40E_VIRTCHNL_OP_VERSION,
I40E_SUCCESS, (uint8_t *)&info, sizeof(info));
 }
-- 
1.7.7.6

[dpdk-dev] [PATCH v2 3/4] net/i40e: parse more VF parameter and configure

[Chen Jing D(Mark)]

When VF requested to configure TX queue, a few parameters are
missed to be configured in PF host. This change have more
fields parsed and configured for TX context.

Signed-off-by: Chen Jing D(Mark) 
---
 drivers/net/i40e/i40e_pf.c |4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/drivers/net/i40e/i40e_pf.c b/drivers/net/i40e/i40e_pf.c
index fd83c16..f12a6aa 100644
--- a/drivers/net/i40e/i40e_pf.c
+++ b/drivers/net/i40e/i40e_pf.c
@@ -400,10 +400,12 @@
 
/* clear the context structure first */
memset(&tx_ctx, 0, sizeof(tx_ctx));
-   tx_ctx.new_context = 1;
tx_ctx.base = txq->dma_ring_addr / I40E_QUEUE_BASE_ADDR_UNIT;
tx_ctx.qlen = txq->ring_len;
tx_ctx.rdylist = rte_le_to_cpu_16(vf->vsi->info.qs_handle[0]);
+   tx_ctx.head_wb_ena = txq->headwb_enabled;
+   tx_ctx.head_wb_addr = txq->dma_headwb_addr;
+
err = i40e_clear_lan_tx_queue_context(hw, abs_queue_id);
if (err != I40E_SUCCESS)
return err;
-- 
1.7.7.6

[dpdk-dev] [PATCH v2 4/4] net/i40e: support Linux VF to configure IRQ link list

[Chen Jing D(Mark)]

i40e PF host only support to work with DPDK VF driver, Linux
VF driver is not supported. This change will enhance in
configuring IRQ link list.

This Change will identify VF client by number of vector
requested. DPDK VF will ask only single one while Linux VF
will request at least 2. It will have different configuration
for different clients. DPDK VF will be configured to link all
queue together, while Linux VF will be configured per request.

Signed-off-by: Chen Jing D(Mark) 
---
 drivers/net/i40e/i40e_pf.c |  152 
 1 files changed, 139 insertions(+), 13 deletions(-)

diff --git a/drivers/net/i40e/i40e_pf.c b/drivers/net/i40e/i40e_pf.c
index f12a6aa..384dfe9 100644
--- a/drivers/net/i40e/i40e_pf.c
+++ b/drivers/net/i40e/i40e_pf.c
@@ -546,13 +546,116 @@
return ret;
 }
 
+static void
+i40e_pf_config_irq_link_list(struct i40e_pf_vf *vf,
+ struct i40e_virtchnl_vector_map *vvm)
+{
+#define BITS_PER_CHAR 8
+   uint64_t linklistmap = 0, tempmap;
+   struct i40e_hw *hw = I40E_PF_TO_HW(vf->pf);
+   uint16_t qid;
+   bool b_first_q = true;
+   enum i40e_queue_type qtype;
+   uint16_t vector_id;
+   uint32_t reg, reg_idx;
+   uint16_t itr_idx = 0, i;
+
+   vector_id = vvm->vector_id;
+   /* setup the head */
+   if (!vector_id)
+   reg_idx = I40E_VPINT_LNKLST0(vf->vf_idx);
+   else
+   reg_idx = I40E_VPINT_LNKLSTN(
+   ((hw->func_caps.num_msix_vectors_vf - 1) * vf->vf_idx)
+   + (vector_id - 1));
+
+   if (vvm->rxq_map == 0 && vvm->txq_map == 0) {
+   I40E_WRITE_REG(hw, reg_idx,
+   I40E_VPINT_LNKLST0_FIRSTQ_INDX_MASK);
+   goto cfg_irq_done;
+   }
+
+   /* sort all rx and tx queues */
+   tempmap = vvm->rxq_map;
+   for (i = 0; i < sizeof(vvm->rxq_map) * BITS_PER_CHAR; i++) {
+   if (tempmap & 0x1)
+   linklistmap |= (1 << (2 * i));
+   tempmap >>= 1;
+   }
+
+   tempmap = vvm->txq_map;
+   for (i = 0; i < sizeof(vvm->txq_map) * BITS_PER_CHAR; i++) {
+   if (tempmap & 0x1)
+   linklistmap |= (1 << (2 * i + 1));
+   tempmap >>= 1;
+   }
+
+   /* Link all rx and tx queues into a chained list */
+   tempmap = linklistmap;
+   i = 0;
+   b_first_q = true;
+   do {
+   if (tempmap & 0x1) {
+   qtype = (enum i40e_queue_type)(i % 2);
+   qid = vf->vsi->base_queue + i / 2;
+   if (b_first_q) {
+   /* This is header */
+   b_first_q = false;
+   reg = ((qtype <<
+   I40E_VPINT_LNKLSTN_FIRSTQ_TYPE_SHIFT)
+   | qid);
+   } else {
+   /* element in the link list */
+   reg = (vector_id) |
+   (qtype << I40E_QINT_RQCTL_NEXTQ_TYPE_SHIFT) |
+   (qid << I40E_QINT_RQCTL_NEXTQ_INDX_SHIFT) |
+   BIT(I40E_QINT_RQCTL_CAUSE_ENA_SHIFT) |
+   (itr_idx << I40E_QINT_RQCTL_ITR_INDX_SHIFT);
+   }
+   I40E_WRITE_REG(hw, reg_idx, reg);
+   /* find next register to program */
+   switch (qtype) {
+   case I40E_QUEUE_TYPE_RX:
+   reg_idx = I40E_QINT_RQCTL(qid);
+   itr_idx = vvm->rxitr_idx;
+   break;
+   case I40E_QUEUE_TYPE_TX:
+   reg_idx = I40E_QINT_TQCTL(qid);
+   itr_idx = vvm->txitr_idx;
+   break;
+   default:
+   break;
+   }
+   }
+   i++;
+   tempmap >>= 1;
+   } while (tempmap);
+
+   /* Terminate the link list */
+   reg = (vector_id) |
+   (0 << I40E_QINT_RQCTL_NEXTQ_TYPE_SHIFT) |
+   (0x7FF << I40E_QINT_RQCTL_NEXTQ_INDX_SHIFT) |
+   BIT(I40E_QINT_RQCTL_CAUSE_ENA_SHIFT) |
+   (itr_idx << I40E_QINT_RQCTL_ITR_INDX_SHIFT);
+   I40E_WRITE_REG(hw, reg_idx, reg);
+
+cfg_irq_done:
+   I40E_WRITE_FLUSH(hw);
+}
+
 static int
 i40e_pf_host_process_cmd_config_irq_map(struct i40e_pf_vf *vf,
uint8_t *msg, uint16_t msglen)
 {
int ret = I40E_SUCCESS;
+   struct i40e_pf *pf = vf->pf;
+   struct i40e_hw *hw = I40E_PF_TO_HW(vf->pf);
struct i40e_virtchnl_irq_map_info *irqmap =
(struct i40e_virtchnl_irq_map_info *)msg;
+   struct i40e_virtchnl_vector_map *map;
+   int i

[dpdk-dev] A question about the function fill_vec_buf

[wangyunjian]

In function fill_vec_buf, it will happen uint32_t cast to uint16_t, when the 
*desc_chain_len is assigned by the len.
This maybe result in data truncation.

static inline int __attribute__((always_inline))
fill_vec_buf(struct virtio_net *dev, struct vhost_virtqueue *vq,
uint32_t avail_idx, uint32_t 
*vec_idx,
struct buf_vector *buf_vec, 
uint16_t *desc_chain_head,
uint16_t *desc_chain_len)   
 --The 
desc_chain_len is defined uint16_t.
{
uint16_t idx = vq->avail->ring[avail_idx & (vq->size - 1)];
uint32_t vec_id = *vec_idx;
uint32_t len= 0;
   
--The len is defined uint32_t.
struct vring_desc *descs = vq->desc;

*desc_chain_head = idx;
...

while (1) {
if (unlikely(vec_id >= BUF_VECTOR_MAX || idx >= 
vq->size))
return -1;

len += descs[idx].len;
buf_vec[vec_id].buf_addr = descs[idx].addr;
buf_vec[vec_id].buf_len  = descs[idx].len;
buf_vec[vec_id].desc_idx = idx;
vec_id++;

if ((descs[idx].flags & VRING_DESC_F_NEXT) == 0)
break;

idx = descs[idx].next;
}

*desc_chain_len = len;  
   --Here, 
uint32_t cast to uint16_t.
*vec_idx = vec_id;

return 0;
}

[dpdk-dev] 17.05 Roadmap

[O'Driscoll, Tim]

Below are the features that we're planning to submit for the 17.05 release. 
We'll submit a patch to update the roadmap page with this info.

It would be good if others are also willing to share their plans so that we can 
build up a complete picture of what's planned for 17.05 and make sure there's 
no duplication.


Cryptodev DES SW PMD: A PMD will be added which implements an optimized 
software version of the DES algorithm.

Cryptodev DOCSIS BPI+: The cryptodev API will be enhanced to enable processing 
of packets according to the Baseline Privacy Interface Plus (BPI+) 
specification described in the Data-over-Cable Service Interface Specification 
(DOCSIS) Security Specification 
(http://www.cablelabs.com/wp-content/uploads/specdocs/CM-SP-SECv3.1-I06-160602.pdf).
 See the RFC (http://dpdk.org/ml/archives/dev/2016-December/052433.html) for 
further details. Support will be added to the QAT PMD for AES and DES, to the 
new software PMD for DES (see previous item), and to the existing AESNI_MB PMD.

Cryptodev Scheduler: This allows packets to be encrypted/decrypted in either SW 
or HW depending on packet size and HW utilization. Reordering will be supported 
so that packet order can be preserved.

Ethernet 32-bit CRC Generation: An optimized x86 library for CRC-32 will be 
added. A CRC may need to be generated for a downstream frame by the DOCSIS MAC 
layer because the CRC may be removed by the NIC or the frame may be modified by 
the MAC for PHS (packet header suppression). 

API to Configure Programmable Devices: More devices are now supporting 
programmable components, for example the Pipeline Personalization Profiles in 
I40E. An API will be added to allow any programmable device to be configured.

I40E Hardware QoS: Hardware QoS will be supported on the I40E. This will 
include Tx bandwidth control (min and max), and Rx Traffic Class assignment.

Configurable Tunnel Filters for I40E: DPDK support will be added for a new I40E 
admin queue which allows configuration of filter types for cloud filters.

Enable MPLS on I40E: MPLSoUDP and MPLSoGRE will be supported for I40E, 
including the new protocols and filtering (RSS and Flow Director), checksum 
offload and packet type recognition.

Software Eventdev Implementation: The libeventdev API is being added in 17.02 
(http://dpdk.org/ml/archives/dev/2016-December/052877.html). A software 
implementation of this API will be added.

New Vhost Device Type: The vhost-user framework will be expanded so that it can 
support additional device types. Support for SCSI will be added initially, but 
block devices and other device types may be added in future.

Abstraction Layer for QoS: An abstraction layer for Quality of Service (QoS) 
hierarchical schedulers will be implemented. The goal of the abstraction layer 
is to provide a simple generic API that is agnostic of the underlying HW, SW or 
mixed HW-SW implementation. See the RFC 
(http://dpdk.org/ml/archives/dev/2016-November/050956.html) for details.

IPFIX Support: The Internet Protocol Flow Information Export (IPFIX) protocol 
provides network administrators with access to IP Flow information. An 
observation point (one for all the interfaces), metering process and an 
exporter process will be implemented. The observation point and metering 
process will be as defined in RFC 5474 (https://tools.ietf.org/html/rfc5474).

Interrupt Mode for Virtio-User: Interrupt mode support will be added for 
virtio-user, which is a virtual device for high performance container 
networking.

Automated VF Processing of PF Reset for I40E: In 16.07, changes were made for 
both IXGBE and I40E to notify a VF when a PF reset occurs. This will be further 
enhanced for I40E so that the driver handles as much of the processing as 
possible, including things like resetting VF ports.

Tx Buffering in Vhost PMD: Overall throughput drops as the number of guests 
increases on a single host.  This is because a Tx burst is spread across 
multiple guests, so the burst size per guest decreases as the number of guests 
increases. Tx buffering will be added which will increase the Tx burst size and 
increase performance.

Consistent PMD Batching Behaviour: PMD Tx batching behavior is not consistent 
between PMDs. Some PMDs have no limit for nb_pkts (besides number of available 
descriptors), some (like vhost) have a limit on the max number of Tx packets 
(32 for vhost), and some define a max burst size and transmit packets in 
multiple bursts up to that size. The application needs to manage these 
differences. To make things easier for application developers, we're 
considering putting the logic for managing this directly into rte_eth_tx_burst.

Re: [dpdk-dev] [RFC] ethdev: abstraction layer for QoS hierarchical scheduler

[Hemant Agrawal]

On 11/30/2016 11:46 PM, Cristian Dumitrescu wrote:

This RFC proposes an ethdev-based abstraction layer for Quality of Service (QoS)
hierarchical scheduler. The goal of the abstraction layer is to provide a simple
generic API that is agnostic of the underlying HW, SW or mixed HW-SW complex
implementation.

Q1: What is the benefit for having an abstraction layer for QoS hierarchical
layer?
A1: There is growing interest in the industry for handling various HW-based,
SW-based or mixed hierarchical scheduler implementations using a unified DPDK
API.

Q2: Which devices are targeted by this abstraction layer?
A2: All current and future devices that expose a hierarchical scheduler feature
under DPDK, including NICs, FPGAs, ASICs, SOCs, SW libraries.

Q3: Which scheduler hierarchies are supported by the API?
A3: Hopefully any scheduler hierarchy can be described and covered by the
current API. Of course, functional correctness, accuracy and performance levels
depend on the specific implementations of this API.

Q4: Why have this abstraction layer into ethdev as opposed to a new type of
device (e.g. scheddev) similar to ethdev, cryptodev, eventdev, etc?
A4: Packets are sent to the Ethernet device using the ethdev API
rte_eth_tx_burst() function, with the hierarchical scheduling taking place
automatically (i.e. no SW intervention) in HW implementations. Basically, the
hierarchical scheduler is done as part of packet TX operation.
The hierarchical scheduler is typically the last stage before packet TX and it
is tightly integrated with the TX stage. The hierarchical scheduler is just
another offload feature of the Ethernet device, which needs to be accommodated
by the ethdev API similar to any other offload feature (such as RSS, DCB,
flow director, etc).
Once the decision to schedule a specific packet has been taken, this packet
cannot be dropped and it has to be sent over the wire as is, otherwise what
takes place on the wire is not what was planned at scheduling time, so the
scheduling is not accurate (Note: there are some devices which allow prepending
headers to the packet after the scheduling stage at the expense of sending
correction requests back to the scheduler, but this only strengthens the bond
between scheduling and TX).


egress QoS can be applied to a physical or a logical network device.
At present the network devices are presented as ethdev in DPDK. Even a 
logical device can also be presented by creating a new ethdev. So it 
seems to be a good idea to associate it with ethdev.




Q5: Given that the packet scheduling takes place automatically for pure HW
implementations, how does packet scheduling take place for poll-mode SW
implementations?
A5: The API provided function rte_sched_run() is designed to take care of this.
For HW implementations, this function typically does nothing. For SW
implementations, this function is typically expected to perform dequeue of
packets from the hierarchical scheduler and their write to Ethernet device TX
queue, periodic flush of any buffers on enqueue-side into the hierarchical
scheduler for burst-oriented implementations, etc.



I think this is *rte_eth_sched_run* in your APIs.

It will be a no-ops for hw, how do you envision it's usages in the 
typical software. e.g. in the l3fwd application,
 - every time you do a rte_eth_tx_burst - there may be locking concern 
here.

 - creating a per port thread to continue doing rte_eth_sched_run
 - call it in one of the existing polling thread for a port.



Q6: Which are the scheduling algorithms supported?
A6: The fundamental scheduling algorithms that are supported are Strict Priority
(SP) and Weighted Fair Queuing (WFQ). The SP and WFQ algorithms are supported at
the level of each node of the scheduling hierarchy, regardless of the node
level/position in the tree. The SP algorithm is used to schedule between sibling
nodes with different priority, while WFQ is used to schedule between groups of
siblings that have the same priority.
Algorithms such as Weighed Round Robin (WRR), byte-level WRR, Deficit WRR
(DWRR), etc are considered approximations of the ideal WFQ and are therefore
assimilated to WFQ, although an associated implementation-dependent accuracy,
performance and resource usage trade-off might exist.

Q7: Which are the supported congestion management algorithms?
A7: Tail drop, head drop and Weighted Random Early Detection (WRED). They are
available for every leaf node in the hierarchy, subject to the specific
implementation supporting them.

We may need to introduce some kind capability APIS. e.g. NXP HW do not 
support headdrop.



Q8: Is traffic shaping supported?
A8: Yes, there are a number of shapers (rate limiters) that can be supported for
each node in the hierarchy (built-in limit is currently set to 4 per node). Each
shaper can be private to a node (used only by that node) or shared between
multiple nodes.



What do you mean by supporting 4 shaper per node? if you need more 
shapers than create new hiera

Re: [dpdk-dev] [PATCH v3 25/29] net/nfp: use eal I/O device memory read/write API

[Alejandro Lucero]

Thanks for that repo. It really makes things easier.

I have run our NFP PMD unit tests using that repo, and this change seems to
work without no problem.

Thanks again.



On Thu, Jan 12, 2017 at 1:40 PM, Jerin Jacob  wrote:

> On Thu, Jan 12, 2017 at 10:53:17AM +, Alejandro Lucero wrote:
> > Hi,
> >
> > I've tried to find out which dpdk repo should I use for testing this
> change
> > with NFP PMD.
> >
> > It seems rte_read/write functions are not with last dpdk main repo, nor
> > with dpdk-net-next.
> >
> > Can someone tell me which repo should I use?
>
> It is based on the dpdk main repo. This patchset has 29 patches. The
> initial
> patches in the series contains the definition of rte_read/write.
>
> I have created a temporary branch in public repo to make other users to
> test the changes without applying all patches.
>
> https://github.com/jerinjacobk/dpdk.git
> branch: rte_io_v3
> It is based on today's dpdk master and this patchset.
>
>

Re: [dpdk-dev] [PATCH v3 25/29] net/nfp: use eal I/O device memory read/write API

[Jerin Jacob]

On Fri, Jan 13, 2017 at 10:49:15AM +, Alejandro Lucero wrote:
> Thanks for that repo. It really makes things easier.
> 
> I have run our NFP PMD unit tests using that repo, and this change seems to
> work without no problem.
> 
> Thanks again.

Thanks Alejandro. Can I add you Acked by on your NFP PMD specific
change.i.e only for this patch([PATCH v3 25/29] net/nfp: use eal I/O device 
memory read/write API)

> 
> 
> 
> On Thu, Jan 12, 2017 at 1:40 PM, Jerin Jacob  > wrote:
> 
> > On Thu, Jan 12, 2017 at 10:53:17AM +, Alejandro Lucero wrote:
> > > Hi,
> > >
> > > I've tried to find out which dpdk repo should I use for testing this
> > change
> > > with NFP PMD.
> > >
> > > It seems rte_read/write functions are not with last dpdk main repo, nor
> > > with dpdk-net-next.
> > >
> > > Can someone tell me which repo should I use?
> >
> > It is based on the dpdk main repo. This patchset has 29 patches. The
> > initial
> > patches in the series contains the definition of rte_read/write.
> >
> > I have created a temporary branch in public repo to make other users to
> > test the changes without applying all patches.
> >
> > https://github.com/jerinjacobk/dpdk.git
> > branch: rte_io_v3
> > It is based on today's dpdk master and this patchset.
> >
> >

[dpdk-dev] [PATCH v7 0/6] Add MACsec offload support for ixgbe

[Tiwei Bie]

This patch set adds the MACsec offload support for ixgbe.
The testpmd is also updated to support MACsec cmds.

v2:
- Update the documents for testpmd;
- Update the release notes;
- Reuse the functions provided by base code;

v3:
- Add the missing parts of MACsec mbuf flag and reorganize the patch set;
- Add an ethdev event type for MACsec;
- Advertise the MACsec offload capabilities based on the mac type;
- Minor fixes and improvements;

v4:
- Reserve bits in mbuf and ethdev for PMD specific API;
- Use the reserved bits in PMD specific API;

v5:
- Add MACsec offload in the NIC feature list;
- Minor improvements on comments;

v6:
- Revert the changes related to the reserved flags;
- Rebase the patch set on the latest branch, xstats code is changed recently;
- Update the feature list when adding the MACsec support for ixgbe;

v7:
- Fix clang build;

Tiwei Bie (6):
  mbuf: add flag for MACsec
  ethdev: add event type for MACsec
  ethdev: add MACsec offload capability flags
  net/ixgbe: add MACsec offload support
  app/testpmd: add MACsec offload commands
  doc: add ixgbe specific APIs

 app/test-pmd/cmdline.c  | 389 ++
 app/test-pmd/macfwd.c   |   2 +
 app/test-pmd/macswap.c  |   2 +
 app/test-pmd/testpmd.h  |   2 +
 app/test-pmd/txonly.c   |   2 +
 doc/guides/nics/features/default.ini|   1 +
 doc/guides/nics/features/ixgbe.ini  |   1 +
 doc/guides/rel_notes/release_17_02.rst  |  10 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  32 ++
 drivers/net/ixgbe/ixgbe_ethdev.c| 482 +++-
 drivers/net/ixgbe/ixgbe_ethdev.h|  45 +++
 drivers/net/ixgbe/ixgbe_rxtx.c  |   3 +
 drivers/net/ixgbe/rte_pmd_ixgbe.h   | 100 ++
 drivers/net/ixgbe/rte_pmd_ixgbe_version.map |  11 +
 lib/librte_ether/rte_ethdev.h   |   3 +
 lib/librte_mbuf/rte_mbuf.c  |   2 +
 lib/librte_mbuf/rte_mbuf.h  |   6 +
 17 files changed, 1088 insertions(+), 5 deletions(-)

-- 
2.7.4

[dpdk-dev] [PATCH v7 2/6] ethdev: add event type for MACsec

[Tiwei Bie]

This commit adds a below event type:

- RTE_ETH_EVENT_MACSEC

This event will occur when the PN counter in a MACsec connection
reaches the exhaustion threshold.

Signed-off-by: Tiwei Bie 
---
 lib/librte_ether/rte_ethdev.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index 1c356c1..d5d2956 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -3188,6 +3188,7 @@ enum rte_eth_event_type {
RTE_ETH_EVENT_INTR_RESET,
/**< reset interrupt event, sent to VF on PF reset */
RTE_ETH_EVENT_VF_MBOX,  /**< message from the VF received by PF */
+   RTE_ETH_EVENT_MACSEC,   /**< MACsec offload related event */
RTE_ETH_EVENT_MAX   /**< max value of this enum */
 };
 
-- 
2.7.4

[dpdk-dev] [PATCH v7 1/6] mbuf: add flag for MACsec

[Tiwei Bie]

Add a new Tx flag in mbuf, that can be set by applications to
enable the MACsec offload for a packet to be transmitted.

Signed-off-by: Tiwei Bie 
Acked-by: Olivier Matz 
---
 doc/guides/rel_notes/release_17_02.rst | 5 +
 lib/librte_mbuf/rte_mbuf.c | 2 ++
 lib/librte_mbuf/rte_mbuf.h | 6 ++
 3 files changed, 13 insertions(+)

diff --git a/doc/guides/rel_notes/release_17_02.rst 
b/doc/guides/rel_notes/release_17_02.rst
index 180af82..3958714 100644
--- a/doc/guides/rel_notes/release_17_02.rst
+++ b/doc/guides/rel_notes/release_17_02.rst
@@ -52,6 +52,11 @@ New Features
   See the :ref:`Generic flow API ` documentation for more
   information.
 
+* **Improved offload support in mbuf.**
+
+  Added a new Tx mbuf flag for MACsec, which can be set by applications
+  to enable the MACsec offload for the packets to be transmitted.
+
 
 Resolved Issues
 ---
diff --git a/lib/librte_mbuf/rte_mbuf.c b/lib/librte_mbuf/rte_mbuf.c
index 63f43c8..72ad91e 100644
--- a/lib/librte_mbuf/rte_mbuf.c
+++ b/lib/librte_mbuf/rte_mbuf.c
@@ -404,6 +404,7 @@ const char *rte_get_tx_ol_flag_name(uint64_t mask)
case PKT_TX_TUNNEL_GRE: return "PKT_TX_TUNNEL_GRE";
case PKT_TX_TUNNEL_IPIP: return "PKT_TX_TUNNEL_IPIP";
case PKT_TX_TUNNEL_GENEVE: return "PKT_TX_TUNNEL_GENEVE";
+   case PKT_TX_MACSEC: return "PKT_TX_MACSEC";
default: return NULL;
}
 }
@@ -434,6 +435,7 @@ rte_get_tx_ol_flag_list(uint64_t mask, char *buf, size_t 
buflen)
  "PKT_TX_TUNNEL_NONE" },
{ PKT_TX_TUNNEL_GENEVE, PKT_TX_TUNNEL_MASK,
  "PKT_TX_TUNNEL_NONE" },
+   { PKT_TX_MACSEC, PKT_TX_MACSEC, NULL },
};
const char *name;
unsigned int i;
diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index 4476d75..ffbb4b3 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -182,6 +182,12 @@ extern "C" {
 /* add new TX flags here */
 
 /**
+ * Offload the MACsec. This flag must be set by the application to enable
+ * this offload feature for a packet to be transmitted.
+ */
+#define PKT_TX_MACSEC(1ULL << 44)
+
+/**
  * Bits 45:48 used for the tunnel type.
  * When doing Tx offload like TSO or checksum, the HW needs to configure the
  * tunnel type into the HW descriptors.
-- 
2.7.4

Re: [dpdk-dev] [PATCH v3 25/29] net/nfp: use eal I/O device memory read/write API

[Alejandro Lucero]

On Fri, Jan 13, 2017 at 10:57 AM, Jerin Jacob <
jerin.ja...@caviumnetworks.com> wrote:

> On Fri, Jan 13, 2017 at 10:49:15AM +, Alejandro Lucero wrote:
> > Thanks for that repo. It really makes things easier.
> >
> > I have run our NFP PMD unit tests using that repo, and this change seems
> to
> > work without no problem.
> >
> > Thanks again.
>
> Thanks Alejandro. Can I add you Acked by on your NFP PMD specific
> change.i.e only for this patch([PATCH v3 25/29] net/nfp: use eal I/O
> device memory read/write API)
>
>
Sure.

Thanks!


> >
> >
> >
> > On Thu, Jan 12, 2017 at 1:40 PM, Jerin Jacob <
> jerin.ja...@caviumnetworks.com
> > > wrote:
> >
> > > On Thu, Jan 12, 2017 at 10:53:17AM +, Alejandro Lucero wrote:
> > > > Hi,
> > > >
> > > > I've tried to find out which dpdk repo should I use for testing this
> > > change
> > > > with NFP PMD.
> > > >
> > > > It seems rte_read/write functions are not with last dpdk main repo,
> nor
> > > > with dpdk-net-next.
> > > >
> > > > Can someone tell me which repo should I use?
> > >
> > > It is based on the dpdk main repo. This patchset has 29 patches. The
> > > initial
> > > patches in the series contains the definition of rte_read/write.
> > >
> > > I have created a temporary branch in public repo to make other users to
> > > test the changes without applying all patches.
> > >
> > > https://github.com/jerinjacobk/dpdk.git
> > > branch: rte_io_v3
> > > It is based on today's dpdk master and this patchset.
> > >
> > >
>

Re: [dpdk-dev] [dpdk-users] IGB_UIO: PCI Resources Management

[Alejandro Lucero]

I completely misread the patch, and I wrongly thought that code was linked
to module removal, but I see this is not about that, but about releasing
the /dev/uio file calling release function, what is done by the kernel when
the process exits.

So yes, the patch avoids the problem I talked about.

However, calling that specific ixgbe driver function will break other
devices relying on igb_uio. What about implementing a notifier chain for
this? The igb_uio code would be agnostic and each interested driver, like
ixgbe or nfp_net, could execute the specific port close code when the
notifier chain triggers.


On Fri, Jan 13, 2017 at 5:33 AM, Tan, Jianfeng 
wrote:

>
>
> > -Original Message-
> > From: Yigit, Ferruh
> > Sent: Friday, January 13, 2017 10:05 AM
> > To: Tan, Jianfeng; Alejandro Lucero
> > Cc: Gregory Etelson; dev; us...@dpdk.org
> > Subject: Re: [dpdk-users] [dpdk-dev] IGB_UIO: PCI Resources Management
> >
> > On 1/13/2017 1:51 AM, Tan, Jianfeng wrote:
> > >
> > >
> > >> -Original Message-
> > >> From: users [mailto:users-boun...@dpdk.org] On Behalf Of Ferruh Yigit
> > >> Sent: Thursday, January 12, 2017 8:22 PM
> > >> To: Alejandro Lucero
> > >> Cc: Gregory Etelson; dev; us...@dpdk.org
> > >> Subject: Re: [dpdk-users] [dpdk-dev] IGB_UIO: PCI Resources
> > Management
> > >>
> > >> On 1/12/2017 12:12 PM, Alejandro Lucero wrote:
> > >>>
> > >>>
> > >>> On Thu, Jan 12, 2017 at 11:55 AM, Ferruh Yigit <
> ferruh.yi...@intel.com
> > >>> > wrote:
> > >>>
> > >>> On 12/9/2016 8:54 AM, Gregory Etelson wrote:
> > >>> > Hello,
> > >>> >
> > >>> > IGB_UIO driver does not close port PCI activities after DPDK
> process
> > >> exits.
> > >>> > DPDK API provides rte_eth_dev_close() to manage port PCI,
> > >>> > but it can be skipped if process receives SIGKILL signal
> > >>>
> > >>> I guess I understand the problem.
> > >>>
> > >>>
> > >>> This is a known problem, but it is not just a UIO problem, and this
> > >>> patch does not solve it, maybe it just solves part of it.
> > >>>
> > >>> In fact, a DPDK program crashing could imply the NIC DMAing after
> that
> > >>> and after that memory was assigned to another program.
> > >>
> > >> Yes.
> > >> Can there be a way to stop NIC DMA, (or prevent it access to mem
> > >> anymore) when app crashes?
> > >> I think that is what this patch is looking for.
> > >
> > > If I understand it correctly, you are looking for this patch?
> > > http://dpdk.org/dev/patchwork/patch/17495/
> > >
> >
> > That is good, thanks Jianfeng, I will check it.
> >
> > btw, patch's current state is rejected, which is by mistake, it seems I
> > confused it with "iomem and ioport mapping" patch, sorry about it, I
> > will update its status immediately.
>
> No problem at all. This patch is rejected as it's based on "iomem and
> ioport mapping" patch. As "iomem and ioport mapping" patch has backward
> compatibility issue, we need to figure out a way to resubmit this patch
> without changing the original "iomem and ioport mapping" in igb_uio.
>
> Thanks,
> Jianfeng
>

Re: [dpdk-dev] [PATCH v6 14/18] net/ixgbe: parse L2 tunnel filter

[Ferruh Yigit]

On 1/13/2017 8:13 AM, Wei Zhao wrote:
> check if the rule is a L2 tunnel rule, and get the L2 tunnel info.
> 
> Signed-off-by: Wei Zhao 
> Signed-off-by: Wenzhuo Lu 
> ---
>  drivers/net/ixgbe/ixgbe_ethdev.c |   3 +-
>  drivers/net/ixgbe/ixgbe_flow.c   | 216 
> +++
>  lib/librte_ether/rte_flow.h  |  48 +
>  3 files changed, 266 insertions(+), 1 deletion(-)

> <...>

Hi Adrien,

Can you please review rte_flow related part of the patch below?

I am OK with rest of the patch, and planning to apply to next-net if you
don't have any objection.

Thanks,
ferruh


>  
> diff --git a/lib/librte_ether/rte_flow.h b/lib/librte_ether/rte_flow.h
> index 98084ac..7142479 100644
> --- a/lib/librte_ether/rte_flow.h
> +++ b/lib/librte_ether/rte_flow.h
> @@ -268,6 +268,20 @@ enum rte_flow_item_type {
>* See struct rte_flow_item_vxlan.
>*/
>   RTE_FLOW_ITEM_TYPE_VXLAN,
> +
> + /**
> +  * Matches a E_TAG header.
> +  *
> +  * See struct rte_flow_item_e_tag.
> +  */
> + RTE_FLOW_ITEM_TYPE_E_TAG,
> +
> + /**
> +  * Matches a NVGRE header.
> +  *
> +  * See struct rte_flow_item_nvgre.
> +  */
> + RTE_FLOW_ITEM_TYPE_NVGRE,
>  };
>  
>  /**
> @@ -454,6 +468,40 @@ struct rte_flow_item_vxlan {
>  };
>  
>  /**
> + * RTE_FLOW_ITEM_TYPE_E_TAG.
> + *
> + * Matches a E-tag header.
> + */
> +struct rte_flow_item_e_tag {
> + uint16_t tpid; /**< Tag protocol identifier (0x893F). */
> + /** E-Tag control information (E-TCI). */
> + /**< E-PCP (3b), E-DEI (1b), ingress E-CID base (12b). */
> + uint16_t epcp_edei_in_ecid_b;
> + /**< Reserved (2b), GRP (2b), E-CID base (12b). */
> + uint16_t rsvd_grp_ecid_b;
> + uint8_t in_ecid_e; /**< Ingress E-CID ext. */
> + uint8_t ecid_e; /**< E-CID ext. */
> +};
> +
> +/**
> + * RTE_FLOW_ITEM_TYPE_NVGRE.
> + *
> + * Matches a NVGRE header.
> + */
> +struct rte_flow_item_nvgre {
> + /**
> +  * Checksum (1b), undefined (1b), key bit (1b), sequence number (1b),
> +  * reserved 0 (9b), version (3b).
> +  *
> +  * \c_k_s_rsvd0_ver must have value 0x2000 according to RFC 7637.
> +  */
> + uint16_t c_k_s_rsvd0_ver;
> + uint16_t protocol; /**< Protocol type (0x6558). */
> + uint8_t tni[3]; /**< Virtual subnet ID. */
> + uint8_t flow_id; /**< Flow ID. */
> +};
> +
> +/**
>   * Matching pattern item definition.
>   *
>   * A pattern is formed by stacking items starting from the lowest protocol
> 

[dpdk-dev] [PATCH v7 2/6] ethdev: add event type for MACsec

[Tiwei Bie]

This commit adds a below event type:

- RTE_ETH_EVENT_MACSEC

This event will occur when the PN counter in a MACsec connection
reaches the exhaustion threshold.

Signed-off-by: Tiwei Bie 
---
 lib/librte_ether/rte_ethdev.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index 1c356c1..d5d2956 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -3188,6 +3188,7 @@ enum rte_eth_event_type {
RTE_ETH_EVENT_INTR_RESET,
/**< reset interrupt event, sent to VF on PF reset */
RTE_ETH_EVENT_VF_MBOX,  /**< message from the VF received by PF */
+   RTE_ETH_EVENT_MACSEC,   /**< MACsec offload related event */
RTE_ETH_EVENT_MAX   /**< max value of this enum */
 };
 
-- 
2.7.4

[dpdk-dev] [PATCH v7 3/6] ethdev: add MACsec offload capability flags

[Tiwei Bie]

If these flags are advertised by a PMD, the NIC supports the MACsec
offload. The incoming MACsec traffics can be offloaded transparently
after the MACsec offload is configured correctly by the application.
And the application can set the PKT_TX_MACSEC flag in mbufs to enable
the MACsec offload for the packets to be transmitted.

Signed-off-by: Tiwei Bie 
---
 lib/librte_ether/rte_ethdev.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index d5d2956..49c073b 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -881,6 +881,7 @@ struct rte_eth_conf {
 #define DEV_RX_OFFLOAD_TCP_LRO 0x0010
 #define DEV_RX_OFFLOAD_QINQ_STRIP  0x0020
 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x0040
+#define DEV_RX_OFFLOAD_MACSEC_STRIP 0x0080
 
 /**
  * TX offload capabilities of a device.
@@ -898,6 +899,7 @@ struct rte_eth_conf {
 #define DEV_TX_OFFLOAD_GRE_TNL_TSO  0x0400/**< Used for tunneling 
packet. */
 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x0800/**< Used for tunneling 
packet. */
 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO   0x1000/**< Used for tunneling 
packet. */
+#define DEV_TX_OFFLOAD_MACSEC_INSERT0x2000
 
 /**
  * Ethernet device information
-- 
2.7.4

[dpdk-dev] [PATCH v7 0/6] Add MACsec offload support for ixgbe

[Tiwei Bie]

This patch set adds the MACsec offload support for ixgbe.
The testpmd is also updated to support MACsec cmds.

v2:
- Update the documents for testpmd;
- Update the release notes;
- Reuse the functions provided by base code;

v3:
- Add the missing parts of MACsec mbuf flag and reorganize the patch set;
- Add an ethdev event type for MACsec;
- Advertise the MACsec offload capabilities based on the mac type;
- Minor fixes and improvements;

v4:
- Reserve bits in mbuf and ethdev for PMD specific API;
- Use the reserved bits in PMD specific API;

v5:
- Add MACsec offload in the NIC feature list;
- Minor improvements on comments;

v6:
- Revert the changes related to the reserved flags;
- Rebase the patch set on the latest branch, xstats code is changed recently;
- Update the feature list when adding the MACsec support for ixgbe;

v7:
- Fix clang build;

Tiwei Bie (6):
  mbuf: add flag for MACsec
  ethdev: add event type for MACsec
  ethdev: add MACsec offload capability flags
  net/ixgbe: add MACsec offload support
  app/testpmd: add MACsec offload commands
  doc: add ixgbe specific APIs

 app/test-pmd/cmdline.c  | 389 ++
 app/test-pmd/macfwd.c   |   2 +
 app/test-pmd/macswap.c  |   2 +
 app/test-pmd/testpmd.h  |   2 +
 app/test-pmd/txonly.c   |   2 +
 doc/guides/nics/features/default.ini|   1 +
 doc/guides/nics/features/ixgbe.ini  |   1 +
 doc/guides/rel_notes/release_17_02.rst  |  10 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  32 ++
 drivers/net/ixgbe/ixgbe_ethdev.c| 482 +++-
 drivers/net/ixgbe/ixgbe_ethdev.h|  45 +++
 drivers/net/ixgbe/ixgbe_rxtx.c  |   3 +
 drivers/net/ixgbe/rte_pmd_ixgbe.h   | 100 ++
 drivers/net/ixgbe/rte_pmd_ixgbe_version.map |  11 +
 lib/librte_ether/rte_ethdev.h   |   3 +
 lib/librte_mbuf/rte_mbuf.c  |   2 +
 lib/librte_mbuf/rte_mbuf.h  |   6 +
 17 files changed, 1088 insertions(+), 5 deletions(-)

-- 
2.7.4

[dpdk-dev] [PATCH v7 1/6] mbuf: add flag for MACsec

[Tiwei Bie]

Add a new Tx flag in mbuf, that can be set by applications to
enable the MACsec offload for a packet to be transmitted.

Signed-off-by: Tiwei Bie 
Acked-by: Olivier Matz 
---
 doc/guides/rel_notes/release_17_02.rst | 5 +
 lib/librte_mbuf/rte_mbuf.c | 2 ++
 lib/librte_mbuf/rte_mbuf.h | 6 ++
 3 files changed, 13 insertions(+)

diff --git a/doc/guides/rel_notes/release_17_02.rst 
b/doc/guides/rel_notes/release_17_02.rst
index 180af82..3958714 100644
--- a/doc/guides/rel_notes/release_17_02.rst
+++ b/doc/guides/rel_notes/release_17_02.rst
@@ -52,6 +52,11 @@ New Features
   See the :ref:`Generic flow API ` documentation for more
   information.
 
+* **Improved offload support in mbuf.**
+
+  Added a new Tx mbuf flag for MACsec, which can be set by applications
+  to enable the MACsec offload for the packets to be transmitted.
+
 
 Resolved Issues
 ---
diff --git a/lib/librte_mbuf/rte_mbuf.c b/lib/librte_mbuf/rte_mbuf.c
index 63f43c8..72ad91e 100644
--- a/lib/librte_mbuf/rte_mbuf.c
+++ b/lib/librte_mbuf/rte_mbuf.c
@@ -404,6 +404,7 @@ const char *rte_get_tx_ol_flag_name(uint64_t mask)
case PKT_TX_TUNNEL_GRE: return "PKT_TX_TUNNEL_GRE";
case PKT_TX_TUNNEL_IPIP: return "PKT_TX_TUNNEL_IPIP";
case PKT_TX_TUNNEL_GENEVE: return "PKT_TX_TUNNEL_GENEVE";
+   case PKT_TX_MACSEC: return "PKT_TX_MACSEC";
default: return NULL;
}
 }
@@ -434,6 +435,7 @@ rte_get_tx_ol_flag_list(uint64_t mask, char *buf, size_t 
buflen)
  "PKT_TX_TUNNEL_NONE" },
{ PKT_TX_TUNNEL_GENEVE, PKT_TX_TUNNEL_MASK,
  "PKT_TX_TUNNEL_NONE" },
+   { PKT_TX_MACSEC, PKT_TX_MACSEC, NULL },
};
const char *name;
unsigned int i;
diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index 4476d75..ffbb4b3 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -182,6 +182,12 @@ extern "C" {
 /* add new TX flags here */
 
 /**
+ * Offload the MACsec. This flag must be set by the application to enable
+ * this offload feature for a packet to be transmitted.
+ */
+#define PKT_TX_MACSEC(1ULL << 44)
+
+/**
  * Bits 45:48 used for the tunnel type.
  * When doing Tx offload like TSO or checksum, the HW needs to configure the
  * tunnel type into the HW descriptors.
-- 
2.7.4

[dpdk-dev] [PATCH v7 6/6] doc: add ixgbe specific APIs

[Tiwei Bie]

Add information about the new ixgbe PMD APIs in the release notes.

Signed-off-by: Tiwei Bie 
Acked-by: John McNamara 
---
 doc/guides/rel_notes/release_17_02.rst | 5 +
 1 file changed, 5 insertions(+)

diff --git a/doc/guides/rel_notes/release_17_02.rst 
b/doc/guides/rel_notes/release_17_02.rst
index 3958714..a3490a5 100644
--- a/doc/guides/rel_notes/release_17_02.rst
+++ b/doc/guides/rel_notes/release_17_02.rst
@@ -52,6 +52,11 @@ New Features
   See the :ref:`Generic flow API ` documentation for more
   information.
 
+* **Added APIs for MACsec offload support to the ixgbe PMD.**
+
+  Six new APIs have been added to the ixgbe PMD for MACsec offload support.
+  The declarations for the APIs can be found in ``rte_pmd_ixgbe.h``.
+
 * **Improved offload support in mbuf.**
 
   Added a new Tx mbuf flag for MACsec, which can be set by applications
-- 
2.7.4

[dpdk-dev] [PATCH v7 4/6] net/ixgbe: add MACsec offload support

[Tiwei Bie]

MACsec (or LinkSec, 802.1AE) is a MAC level encryption/authentication
scheme defined in IEEE 802.1AE that uses symmetric cryptography.
This commit adds the MACsec offload support for ixgbe.

Signed-off-by: Tiwei Bie 
Acked-by: Wenzhuo Lu 
---
 doc/guides/nics/features/default.ini|   1 +
 doc/guides/nics/features/ixgbe.ini  |   1 +
 drivers/net/ixgbe/ixgbe_ethdev.c| 482 +++-
 drivers/net/ixgbe/ixgbe_ethdev.h|  45 +++
 drivers/net/ixgbe/ixgbe_rxtx.c  |   3 +
 drivers/net/ixgbe/rte_pmd_ixgbe.h   | 100 ++
 drivers/net/ixgbe/rte_pmd_ixgbe_version.map |  11 +
 7 files changed, 638 insertions(+), 5 deletions(-)

diff --git a/doc/guides/nics/features/default.ini 
b/doc/guides/nics/features/default.ini
index f1bf9bf..c412d6f 100644
--- a/doc/guides/nics/features/default.ini
+++ b/doc/guides/nics/features/default.ini
@@ -43,6 +43,7 @@ VLAN offload =
 QinQ offload =
 L3 checksum offload  =
 L4 checksum offload  =
+MACsec offload   =
 Inner L3 checksum=
 Inner L4 checksum=
 Packet type parsing  =
diff --git a/doc/guides/nics/features/ixgbe.ini 
b/doc/guides/nics/features/ixgbe.ini
index 4a5667f..24c02fb 100644
--- a/doc/guides/nics/features/ixgbe.ini
+++ b/doc/guides/nics/features/ixgbe.ini
@@ -36,6 +36,7 @@ VLAN offload = Y
 QinQ offload = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
+MACsec offload   = Y
 Inner L3 checksum= Y
 Inner L4 checksum= Y
 Packet type parsing  = Y
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index b7415e8..08f4449 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -231,6 +231,7 @@ static int ixgbe_dev_rss_reta_query(struct rte_eth_dev *dev,
uint16_t reta_size);
 static void ixgbe_dev_link_status_print(struct rte_eth_dev *dev);
 static int ixgbe_dev_lsc_interrupt_setup(struct rte_eth_dev *dev);
+static int ixgbe_dev_macsec_interrupt_setup(struct rte_eth_dev *dev);
 static int ixgbe_dev_rxq_interrupt_setup(struct rte_eth_dev *dev);
 static int ixgbe_dev_interrupt_get_status(struct rte_eth_dev *dev);
 static int ixgbe_dev_interrupt_action(struct rte_eth_dev *dev,
@@ -747,6 +748,51 @@ static const struct rte_ixgbe_xstats_name_off 
rte_ixgbe_stats_strings[] = {
 #define IXGBE_NB_HW_STATS (sizeof(rte_ixgbe_stats_strings) / \
   sizeof(rte_ixgbe_stats_strings[0]))
 
+/* MACsec statistics */
+static const struct rte_ixgbe_xstats_name_off rte_ixgbe_macsec_strings[] = {
+   {"out_pkts_untagged", offsetof(struct ixgbe_macsec_stats,
+   out_pkts_untagged)},
+   {"out_pkts_encrypted", offsetof(struct ixgbe_macsec_stats,
+   out_pkts_encrypted)},
+   {"out_pkts_protected", offsetof(struct ixgbe_macsec_stats,
+   out_pkts_protected)},
+   {"out_octets_encrypted", offsetof(struct ixgbe_macsec_stats,
+   out_octets_encrypted)},
+   {"out_octets_protected", offsetof(struct ixgbe_macsec_stats,
+   out_octets_protected)},
+   {"in_pkts_untagged", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_untagged)},
+   {"in_pkts_badtag", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_badtag)},
+   {"in_pkts_nosci", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_nosci)},
+   {"in_pkts_unknownsci", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_unknownsci)},
+   {"in_octets_decrypted", offsetof(struct ixgbe_macsec_stats,
+   in_octets_decrypted)},
+   {"in_octets_validated", offsetof(struct ixgbe_macsec_stats,
+   in_octets_validated)},
+   {"in_pkts_unchecked", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_unchecked)},
+   {"in_pkts_delayed", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_delayed)},
+   {"in_pkts_late", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_late)},
+   {"in_pkts_ok", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_ok)},
+   {"in_pkts_invalid", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_invalid)},
+   {"in_pkts_notvalid", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_notvalid)},
+   {"in_pkts_unusedsa", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_unusedsa)},
+   {"in_pkts_notusingsa", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_notusingsa)},
+};
+
+#define IXGBE_NB_MACSEC_STATS (sizeof(rte_ixgbe_macsec_strings) / \
+  sizeof(rte_ixgbe_macsec_strings[0]))
+
 /* Per-queue statistics */
 static const struct rte_ixgbe_xstats_name_off rte_ixgbe_rxq_strings[] = {
{"mbuf_allocation_errors", offsetof(struct ixgbe_hw_stats, rnbc)},
@@ -2371,6 +2417,7 @@ ixgbe_dev_start(struct rte_eth_dev *dev)
/* check if lsc interrupt is enabled */
if (dev->data->dev_co

[dpdk-dev] [PATCH v7 5/6] app/testpmd: add MACsec offload commands

[Tiwei Bie]

Below MACsec offload commands are added:

- set macsec offload  on encrypt on|off replay-protect on|off
- set macsec offload  off
- set macsec sc tx|rx   
- set macsec sa tx|rx 

Also update the testpmd user guide.

Signed-off-by: Tiwei Bie 
Acked-by: Wenzhuo Lu 
---
 app/test-pmd/cmdline.c  | 389 
 app/test-pmd/macfwd.c   |   2 +
 app/test-pmd/macswap.c  |   2 +
 app/test-pmd/testpmd.h  |   2 +
 app/test-pmd/txonly.c   |   2 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  32 +++
 6 files changed, 429 insertions(+)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index f768b6b..1bf42ba 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -275,6 +275,18 @@ static void cmd_help_long_parsed(void *parsed_result,
 
"set vf mac antispoof (port_id) (vf_id) (on|off).\n"
"Set MAC antispoof for a VF from the PF.\n\n"
+
+   "set macsec offload (port_id) on encrypt (on|off) 
replay-protect (on|off)\n"
+   "Enable MACsec offload.\n\n"
+
+   "set macsec offload (port_id) off\n"
+   "Disable MACsec offload.\n\n"
+
+   "set macsec sc (tx|rx) (port_id) (mac) (pi)\n"
+   "Configure MACsec secure connection (SC).\n\n"
+
+   "set macsec sa (tx|rx) (port_id) (idx) (an) (pn) 
(key)\n"
+   "Configure MACsec secure association (SA).\n\n"
 #endif
 
"vlan set strip (on|off) (port_id)\n"
@@ -11488,6 +11500,379 @@ cmdline_parse_inst_t cmd_set_vf_mac_addr = {
NULL,
},
 };
+
+/* MACsec configuration */
+
+/* Common result structure for MACsec offload enable */
+struct cmd_macsec_offload_on_result {
+   cmdline_fixed_string_t set;
+   cmdline_fixed_string_t macsec;
+   cmdline_fixed_string_t offload;
+   uint8_t port_id;
+   cmdline_fixed_string_t on;
+   cmdline_fixed_string_t encrypt;
+   cmdline_fixed_string_t en_on_off;
+   cmdline_fixed_string_t replay_protect;
+   cmdline_fixed_string_t rp_on_off;
+};
+
+/* Common CLI fields for MACsec offload disable */
+cmdline_parse_token_string_t cmd_macsec_offload_on_set =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+set, "set");
+cmdline_parse_token_string_t cmd_macsec_offload_on_macsec =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+macsec, "macsec");
+cmdline_parse_token_string_t cmd_macsec_offload_on_offload =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+offload, "offload");
+cmdline_parse_token_num_t cmd_macsec_offload_on_port_id =
+   TOKEN_NUM_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+port_id, UINT8);
+cmdline_parse_token_string_t cmd_macsec_offload_on_on =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+on, "on");
+cmdline_parse_token_string_t cmd_macsec_offload_on_encrypt =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+encrypt, "encrypt");
+cmdline_parse_token_string_t cmd_macsec_offload_on_en_on_off =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+en_on_off, "on#off");
+cmdline_parse_token_string_t cmd_macsec_offload_on_replay_protect =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+replay_protect, "replay-protect");
+cmdline_parse_token_string_t cmd_macsec_offload_on_rp_on_off =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+rp_on_off, "on#off");
+
+static void
+cmd_set_macsec_offload_on_parsed(
+   void *parsed_result,
+   __attribute__((unused)) struct cmdline *cl,
+   __attribute__((unused)) void *data)
+{
+   struct cmd_macsec_offload_on_result *res = parsed_result;
+   int ret;
+   portid_t port_id = res->port_id;
+   int en = (strcmp(res->en_on_off, "on") == 0) ? 1 : 0;
+   int rp = (strcmp(res->rp_on_off, "on") == 0) ? 1 : 0;
+
+   if (port_id_is_invalid(port_id, ENABLED_WARN))
+   return;
+
+   ports[port_id].tx_ol_flags |= TESTPMD_TX_OFFLOAD_MACSEC;
+   ret = rte_pmd_ixgbe_macsec_enable(port_id, en, rp);
+
+   switch (ret) {
+   case 0:
+   break;
+   case -ENODEV:
+   printf("invalid port_id %d\n", port_id);
+   break;
+   default:
+   printf("programming error: (%s)\n", strerror(-ret));
+   }
+}
+
+cmdline_parse_inst_t cmd_set_macsec_offload_on = {
+   .f = cmd_set_macsec_offload_on_p

Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device memory read/write API

[Ferruh Yigit]

On 1/13/2017 8:17 AM, Jerin Jacob wrote:
> On Thu, Jan 12, 2017 at 07:09:22PM +, Ferruh Yigit wrote:
>> Hi Jerin,
>>
>> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
>> <...>
>>
>>> +#include 
>>> +
>>>  /* CSR write macro */
>>> -#define ADF_CSR_WR(csrAddr, csrOffset, val) \
>>> -   (void)((*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)) \
>>> -   = (val)))
>>> +#define ADF_CSR_WR(csrAddr, csrOffset, val)\
>>> +   rte_write32(val, (((uint8_t *)csrAddr) + csrOffset))
>>
>> For IA, this update introduces an extra compiler barrier (rte_io_wmb()),
>> which is indeed not a must, is this correct?
> 
> AFAIK, Compiler barrier is required for IA. I am not an IA expert, if
> someone thinks it needs to changed then I can fix it in following commit
> in this patch series by making rte_io_wmb() and rte_io_rmb() as empty.
> 
> Let me know.
> 
> AFAIK, Linux kernel code has a barrier in readl/writel for IA.
> 
> Typically we don't use any non relaxed versions in fast path.In fast
> typically all the drivers has explicit write barrier for doorbell write
> and followed by a relaxed version of write. IMO, In any event, it won't
> generate performance regression.
> 
> [dpdk-master] $ git show
> 70c343bdc8c33a51a9db23cd58122bdfc120a58f
> commit 70c343bdc8c33a51a9db23cd58122bdfc120a58f
> Author: Jerin Jacob 
> Date:   Mon Dec 5 06:36:49 2016 +0530
> 
> eal/x86: define I/O device memory barriers for IA
> 
> The patch does not provide any functional change for IA.
> I/O barriers are mapped to existing smp barriers.
> 
> CC: Bruce Richardson 
> CC: Konstantin Ananyev 
> Signed-off-by: Jerin Jacob 
> 
> diff --git a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> index 00b1cdf..4eac666 100644
> --- a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> +++ b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> @@ -61,6 +61,12 @@ extern "C" {
>  
>  #define rte_smp_rmb() rte_compiler_barrier()
>  
> +#define rte_io_mb() rte_mb()
> +
> +#define rte_io_wmb() rte_compiler_barrier()
> +
> +#define rte_io_rmb() rte_compiler_barrier()
> +
>  /*- 16 bit atomic operations
>  * -*/
>  
>  #ifndef RTE_FORCE_INTRINSICS
> 
>>
>> If so, does it make sense to override these functions for x86, and make
>> rte_writeX = rte_writeX_relaxed
>> rte_readX = rte_readX_relaxed
>>
>>>  
>>>  /* CSR read macro */
>>> -#define ADF_CSR_RD(csrAddr, csrOffset) \
>>> -   (*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)))
>>> +#define ADF_CSR_RD(csrAddr, csrOffset) \
>>> +   rte_read32uint8_t *)csrAddr) + csrOffset))
>>
>> This patchset both introduces new rte_readX/rte_writeX functions, also
>> applies them into drivers.
>>
>> While applying them, it changes the behavior.
>> Like above code was doing a read, but after update it does read and
>> read_memory_barrier.
>>
>> What do you think this patchset updates usage in a manner that keeps
>> behavior exact same. Like using rte_read32_relaxed for this case.
>> And doing architecture related updates in a different patchset?
> 
> Need to use rte_read32 at this commit otherwise it will break for ARM.
> That's was all point for this patchset.

Why it breaks the ARM, is it because rte_*mb() updated for ARM in this
patchset (patch 7/29) ?

I believe it is good to make these modifications in two phase:
- First replace old usage with rte_readX/rte_writeX while keeping exact
same behavior

- Second, do architecture specific changes. Both in eal and drivers
level if required.

Thanks,
ferruh

> For performance regression, we can always verify by taking delta
> between this changeset and the previous changeset. If you think, I need
> to make rte_io_wmb()/rte_io_rmb() as empty for IA then I could do that
> as well.
> 
> 
>>
>> This both makes easy to see architecture specific updates, and makes
>> easy to trace any possible performance issues by this patchset.
>>
>>>  
>>>  #define ADF_BANK_INT_SRC_SEL_MASK_0 0x444CUL
>>>  #define ADF_BANK_INT_SRC_SEL_MASK_X 0xUL
>>>
>>

[dpdk-dev] [PATCH v7 3/6] ethdev: add MACsec offload capability flags

[Tiwei Bie]

If these flags are advertised by a PMD, the NIC supports the MACsec
offload. The incoming MACsec traffics can be offloaded transparently
after the MACsec offload is configured correctly by the application.
And the application can set the PKT_TX_MACSEC flag in mbufs to enable
the MACsec offload for the packets to be transmitted.

Signed-off-by: Tiwei Bie 
---
 lib/librte_ether/rte_ethdev.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index d5d2956..49c073b 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -881,6 +881,7 @@ struct rte_eth_conf {
 #define DEV_RX_OFFLOAD_TCP_LRO 0x0010
 #define DEV_RX_OFFLOAD_QINQ_STRIP  0x0020
 #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x0040
+#define DEV_RX_OFFLOAD_MACSEC_STRIP 0x0080
 
 /**
  * TX offload capabilities of a device.
@@ -898,6 +899,7 @@ struct rte_eth_conf {
 #define DEV_TX_OFFLOAD_GRE_TNL_TSO  0x0400/**< Used for tunneling 
packet. */
 #define DEV_TX_OFFLOAD_IPIP_TNL_TSO 0x0800/**< Used for tunneling 
packet. */
 #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO   0x1000/**< Used for tunneling 
packet. */
+#define DEV_TX_OFFLOAD_MACSEC_INSERT0x2000
 
 /**
  * Ethernet device information
-- 
2.7.4

Re: [dpdk-dev] [PATCH] net/nfp: fix Vlan offload flags check

[Alejandro Lucero]

Acked.

Thanks!

On Thu, Jan 12, 2017 at 1:04 PM, Olivier Matz 
wrote:

> Fix typo when checking that no Vlan offload flags are passed at port
> initialization.
>
> By the way, also fix a typo in the log.
>
> Fixes: d4a27a3b092a ("nfp: add basic features")
>
> Signed-off-by: Olivier Matz 
> ---
>  drivers/net/nfp/nfp_net.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
> index 65ba09f..ed3c9b8 100644
> --- a/drivers/net/nfp/nfp_net.c
> +++ b/drivers/net/nfp/nfp_net.c
> @@ -2174,8 +2174,8 @@ nfp_net_vlan_offload_set(struct rte_eth_dev *dev,
> int mask)
> new_ctrl = 0;
>
> if ((mask & ETH_VLAN_FILTER_OFFLOAD) ||
> -   (mask & ETH_VLAN_FILTER_OFFLOAD))
> -   RTE_LOG(INFO, PMD, "Not support for
> ETH_VLAN_FILTER_OFFLOAD or"
> +   (mask & ETH_VLAN_FILTER_EXTEND))
> +   RTE_LOG(INFO, PMD, "No support for ETH_VLAN_FILTER_OFFLOAD
> or"
> " ETH_VLAN_FILTER_EXTEND");
>
> /* Enable vlan strip if it is not configured yet */
> --
> 2.8.1
>
>

[dpdk-dev] [PATCH v7 6/6] doc: add ixgbe specific APIs

[Tiwei Bie]

Add information about the new ixgbe PMD APIs in the release notes.

Signed-off-by: Tiwei Bie 
Acked-by: John McNamara 
---
 doc/guides/rel_notes/release_17_02.rst | 5 +
 1 file changed, 5 insertions(+)

diff --git a/doc/guides/rel_notes/release_17_02.rst 
b/doc/guides/rel_notes/release_17_02.rst
index 3958714..a3490a5 100644
--- a/doc/guides/rel_notes/release_17_02.rst
+++ b/doc/guides/rel_notes/release_17_02.rst
@@ -52,6 +52,11 @@ New Features
   See the :ref:`Generic flow API ` documentation for more
   information.
 
+* **Added APIs for MACsec offload support to the ixgbe PMD.**
+
+  Six new APIs have been added to the ixgbe PMD for MACsec offload support.
+  The declarations for the APIs can be found in ``rte_pmd_ixgbe.h``.
+
 * **Improved offload support in mbuf.**
 
   Added a new Tx mbuf flag for MACsec, which can be set by applications
-- 
2.7.4

[dpdk-dev] [PATCH v7 5/6] app/testpmd: add MACsec offload commands

[Tiwei Bie]

Below MACsec offload commands are added:

- set macsec offload  on encrypt on|off replay-protect on|off
- set macsec offload  off
- set macsec sc tx|rx   
- set macsec sa tx|rx 

Also update the testpmd user guide.

Signed-off-by: Tiwei Bie 
Acked-by: Wenzhuo Lu 
---
 app/test-pmd/cmdline.c  | 389 
 app/test-pmd/macfwd.c   |   2 +
 app/test-pmd/macswap.c  |   2 +
 app/test-pmd/testpmd.h  |   2 +
 app/test-pmd/txonly.c   |   2 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  32 +++
 6 files changed, 429 insertions(+)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index f768b6b..1bf42ba 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -275,6 +275,18 @@ static void cmd_help_long_parsed(void *parsed_result,
 
"set vf mac antispoof (port_id) (vf_id) (on|off).\n"
"Set MAC antispoof for a VF from the PF.\n\n"
+
+   "set macsec offload (port_id) on encrypt (on|off) 
replay-protect (on|off)\n"
+   "Enable MACsec offload.\n\n"
+
+   "set macsec offload (port_id) off\n"
+   "Disable MACsec offload.\n\n"
+
+   "set macsec sc (tx|rx) (port_id) (mac) (pi)\n"
+   "Configure MACsec secure connection (SC).\n\n"
+
+   "set macsec sa (tx|rx) (port_id) (idx) (an) (pn) 
(key)\n"
+   "Configure MACsec secure association (SA).\n\n"
 #endif
 
"vlan set strip (on|off) (port_id)\n"
@@ -11488,6 +11500,379 @@ cmdline_parse_inst_t cmd_set_vf_mac_addr = {
NULL,
},
 };
+
+/* MACsec configuration */
+
+/* Common result structure for MACsec offload enable */
+struct cmd_macsec_offload_on_result {
+   cmdline_fixed_string_t set;
+   cmdline_fixed_string_t macsec;
+   cmdline_fixed_string_t offload;
+   uint8_t port_id;
+   cmdline_fixed_string_t on;
+   cmdline_fixed_string_t encrypt;
+   cmdline_fixed_string_t en_on_off;
+   cmdline_fixed_string_t replay_protect;
+   cmdline_fixed_string_t rp_on_off;
+};
+
+/* Common CLI fields for MACsec offload disable */
+cmdline_parse_token_string_t cmd_macsec_offload_on_set =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+set, "set");
+cmdline_parse_token_string_t cmd_macsec_offload_on_macsec =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+macsec, "macsec");
+cmdline_parse_token_string_t cmd_macsec_offload_on_offload =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+offload, "offload");
+cmdline_parse_token_num_t cmd_macsec_offload_on_port_id =
+   TOKEN_NUM_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+port_id, UINT8);
+cmdline_parse_token_string_t cmd_macsec_offload_on_on =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+on, "on");
+cmdline_parse_token_string_t cmd_macsec_offload_on_encrypt =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+encrypt, "encrypt");
+cmdline_parse_token_string_t cmd_macsec_offload_on_en_on_off =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+en_on_off, "on#off");
+cmdline_parse_token_string_t cmd_macsec_offload_on_replay_protect =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+replay_protect, "replay-protect");
+cmdline_parse_token_string_t cmd_macsec_offload_on_rp_on_off =
+   TOKEN_STRING_INITIALIZER
+   (struct cmd_macsec_offload_on_result,
+rp_on_off, "on#off");
+
+static void
+cmd_set_macsec_offload_on_parsed(
+   void *parsed_result,
+   __attribute__((unused)) struct cmdline *cl,
+   __attribute__((unused)) void *data)
+{
+   struct cmd_macsec_offload_on_result *res = parsed_result;
+   int ret;
+   portid_t port_id = res->port_id;
+   int en = (strcmp(res->en_on_off, "on") == 0) ? 1 : 0;
+   int rp = (strcmp(res->rp_on_off, "on") == 0) ? 1 : 0;
+
+   if (port_id_is_invalid(port_id, ENABLED_WARN))
+   return;
+
+   ports[port_id].tx_ol_flags |= TESTPMD_TX_OFFLOAD_MACSEC;
+   ret = rte_pmd_ixgbe_macsec_enable(port_id, en, rp);
+
+   switch (ret) {
+   case 0:
+   break;
+   case -ENODEV:
+   printf("invalid port_id %d\n", port_id);
+   break;
+   default:
+   printf("programming error: (%s)\n", strerror(-ret));
+   }
+}
+
+cmdline_parse_inst_t cmd_set_macsec_offload_on = {
+   .f = cmd_set_macsec_offload_on_p

[dpdk-dev] [PATCH v7 4/6] net/ixgbe: add MACsec offload support

[Tiwei Bie]

MACsec (or LinkSec, 802.1AE) is a MAC level encryption/authentication
scheme defined in IEEE 802.1AE that uses symmetric cryptography.
This commit adds the MACsec offload support for ixgbe.

Signed-off-by: Tiwei Bie 
Acked-by: Wenzhuo Lu 
---
 doc/guides/nics/features/default.ini|   1 +
 doc/guides/nics/features/ixgbe.ini  |   1 +
 drivers/net/ixgbe/ixgbe_ethdev.c| 482 +++-
 drivers/net/ixgbe/ixgbe_ethdev.h|  45 +++
 drivers/net/ixgbe/ixgbe_rxtx.c  |   3 +
 drivers/net/ixgbe/rte_pmd_ixgbe.h   | 100 ++
 drivers/net/ixgbe/rte_pmd_ixgbe_version.map |  11 +
 7 files changed, 638 insertions(+), 5 deletions(-)

diff --git a/doc/guides/nics/features/default.ini 
b/doc/guides/nics/features/default.ini
index f1bf9bf..c412d6f 100644
--- a/doc/guides/nics/features/default.ini
+++ b/doc/guides/nics/features/default.ini
@@ -43,6 +43,7 @@ VLAN offload =
 QinQ offload =
 L3 checksum offload  =
 L4 checksum offload  =
+MACsec offload   =
 Inner L3 checksum=
 Inner L4 checksum=
 Packet type parsing  =
diff --git a/doc/guides/nics/features/ixgbe.ini 
b/doc/guides/nics/features/ixgbe.ini
index 4a5667f..24c02fb 100644
--- a/doc/guides/nics/features/ixgbe.ini
+++ b/doc/guides/nics/features/ixgbe.ini
@@ -36,6 +36,7 @@ VLAN offload = Y
 QinQ offload = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
+MACsec offload   = Y
 Inner L3 checksum= Y
 Inner L4 checksum= Y
 Packet type parsing  = Y
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index b7415e8..08f4449 100644
--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c
@@ -231,6 +231,7 @@ static int ixgbe_dev_rss_reta_query(struct rte_eth_dev *dev,
uint16_t reta_size);
 static void ixgbe_dev_link_status_print(struct rte_eth_dev *dev);
 static int ixgbe_dev_lsc_interrupt_setup(struct rte_eth_dev *dev);
+static int ixgbe_dev_macsec_interrupt_setup(struct rte_eth_dev *dev);
 static int ixgbe_dev_rxq_interrupt_setup(struct rte_eth_dev *dev);
 static int ixgbe_dev_interrupt_get_status(struct rte_eth_dev *dev);
 static int ixgbe_dev_interrupt_action(struct rte_eth_dev *dev,
@@ -747,6 +748,51 @@ static const struct rte_ixgbe_xstats_name_off 
rte_ixgbe_stats_strings[] = {
 #define IXGBE_NB_HW_STATS (sizeof(rte_ixgbe_stats_strings) / \
   sizeof(rte_ixgbe_stats_strings[0]))
 
+/* MACsec statistics */
+static const struct rte_ixgbe_xstats_name_off rte_ixgbe_macsec_strings[] = {
+   {"out_pkts_untagged", offsetof(struct ixgbe_macsec_stats,
+   out_pkts_untagged)},
+   {"out_pkts_encrypted", offsetof(struct ixgbe_macsec_stats,
+   out_pkts_encrypted)},
+   {"out_pkts_protected", offsetof(struct ixgbe_macsec_stats,
+   out_pkts_protected)},
+   {"out_octets_encrypted", offsetof(struct ixgbe_macsec_stats,
+   out_octets_encrypted)},
+   {"out_octets_protected", offsetof(struct ixgbe_macsec_stats,
+   out_octets_protected)},
+   {"in_pkts_untagged", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_untagged)},
+   {"in_pkts_badtag", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_badtag)},
+   {"in_pkts_nosci", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_nosci)},
+   {"in_pkts_unknownsci", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_unknownsci)},
+   {"in_octets_decrypted", offsetof(struct ixgbe_macsec_stats,
+   in_octets_decrypted)},
+   {"in_octets_validated", offsetof(struct ixgbe_macsec_stats,
+   in_octets_validated)},
+   {"in_pkts_unchecked", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_unchecked)},
+   {"in_pkts_delayed", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_delayed)},
+   {"in_pkts_late", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_late)},
+   {"in_pkts_ok", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_ok)},
+   {"in_pkts_invalid", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_invalid)},
+   {"in_pkts_notvalid", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_notvalid)},
+   {"in_pkts_unusedsa", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_unusedsa)},
+   {"in_pkts_notusingsa", offsetof(struct ixgbe_macsec_stats,
+   in_pkts_notusingsa)},
+};
+
+#define IXGBE_NB_MACSEC_STATS (sizeof(rte_ixgbe_macsec_strings) / \
+  sizeof(rte_ixgbe_macsec_strings[0]))
+
 /* Per-queue statistics */
 static const struct rte_ixgbe_xstats_name_off rte_ixgbe_rxq_strings[] = {
{"mbuf_allocation_errors", offsetof(struct ixgbe_hw_stats, rnbc)},
@@ -2371,6 +2417,7 @@ ixgbe_dev_start(struct rte_eth_dev *dev)
/* check if lsc interrupt is enabled */
if (dev->data->dev_co

Re: [dpdk-dev] [PATCH v3 18/29] net/cxgbe: use eal I/O device memory read/write API

[Ferruh Yigit]

On 1/13/2017 8:37 AM, Jerin Jacob wrote:
> On Thu, Jan 12, 2017 at 07:12:33PM +, Ferruh Yigit wrote:
>> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
>> <...>
>>
>>>  
>>> -#define CXGBE_PCI_REG_WRITE(reg, value) ({ \
>>> -   CXGBE_PCI_REG((reg)) = (value); })
>>> +#define CXGBE_PCI_REG_WRITE(reg, value) rte_write32((value), (reg))
>>
>> Almost all (if not all) PMD write macros' argument order is like
>> write(address, value), but rte_writeX has rte_writex(value, address)
>>
>> What is the common usage for this kind of function?
> 
> Arguments order has been taken from Linux kernel readl/writel syntax.
> 
>> What do you think reverting argument order?
> 
> OMG :-). If it worth it then we can consider. IMHO, let it be in kernel
> syntax so that it will easy to port drivers from Linux kernel.

Fair enough.

> 
>>
>> As a similar example, dpdk ether_addr_copy(src, dst) function argument
>> order is revers according common usage, and keeps confusing people.
>>
>> <...>
>>

Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device memory read/write API

[Bruce Richardson]

On Fri, Jan 13, 2017 at 11:40:06AM +, Ferruh Yigit wrote:
> On 1/13/2017 8:17 AM, Jerin Jacob wrote:
> > On Thu, Jan 12, 2017 at 07:09:22PM +, Ferruh Yigit wrote:
> >> Hi Jerin,
> >>
> >> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> >> <...>
> >>
> >>> +#include 
> >>> +
> >>>  /* CSR write macro */
> >>> -#define ADF_CSR_WR(csrAddr, csrOffset, val) \
> >>> - (void)((*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)) \
> >>> - = (val)))
> >>> +#define ADF_CSR_WR(csrAddr, csrOffset, val)  \
> >>> + rte_write32(val, (((uint8_t *)csrAddr) + csrOffset))
> >>
> >> For IA, this update introduces an extra compiler barrier (rte_io_wmb()),
> >> which is indeed not a must, is this correct?
> > 
> > AFAIK, Compiler barrier is required for IA. I am not an IA expert, if
> > someone thinks it needs to changed then I can fix it in following commit
> > in this patch series by making rte_io_wmb() and rte_io_rmb() as empty.
> > 
> > Let me know.
> 
> I don't know, but what I know is this was working for IA without
> compiler barrier before.
> 
> Bruce or Konstantin can help here.
> 
I think having a compiler barrier is safer. If all data being written
before the actual register write to the device is volatile, none is
needed, but also in that case, the compiler barrier should have no
effect. If some of the data is not volatile, then a barrier is needed
for correctness. IMHO, unless there is a performance regression by doing
so, I think having the IO register writes have compiler barriers on IA
is a good thing. It should save individual drivers from having to worry
about the barriers themselves in most cases.

Regards,
/Bruce

Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device memory read/write API

[Ferruh Yigit]

On 1/13/2017 8:17 AM, Jerin Jacob wrote:
> On Thu, Jan 12, 2017 at 07:09:22PM +, Ferruh Yigit wrote:
>> Hi Jerin,
>>
>> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
>> <...>
>>
>>> +#include 
>>> +
>>>  /* CSR write macro */
>>> -#define ADF_CSR_WR(csrAddr, csrOffset, val) \
>>> -   (void)((*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)) \
>>> -   = (val)))
>>> +#define ADF_CSR_WR(csrAddr, csrOffset, val)\
>>> +   rte_write32(val, (((uint8_t *)csrAddr) + csrOffset))
>>
>> For IA, this update introduces an extra compiler barrier (rte_io_wmb()),
>> which is indeed not a must, is this correct?
> 
> AFAIK, Compiler barrier is required for IA. I am not an IA expert, if
> someone thinks it needs to changed then I can fix it in following commit
> in this patch series by making rte_io_wmb() and rte_io_rmb() as empty.
> 
> Let me know.

I don't know, but what I know is this was working for IA without
compiler barrier before.

Bruce or Konstantin can help here.

> 
> AFAIK, Linux kernel code has a barrier in readl/writel for IA.
> 
> Typically we don't use any non relaxed versions in fast path.In fast
> typically all the drivers has explicit write barrier for doorbell write
> and followed by a relaxed version of write. IMO, In any event, it won't
> generate performance regression.
> 
> [dpdk-master] $ git show
> 70c343bdc8c33a51a9db23cd58122bdfc120a58f
> commit 70c343bdc8c33a51a9db23cd58122bdfc120a58f
> Author: Jerin Jacob 
> Date:   Mon Dec 5 06:36:49 2016 +0530
> 
> eal/x86: define I/O device memory barriers for IA
> 
> The patch does not provide any functional change for IA.
> I/O barriers are mapped to existing smp barriers.
> 
> CC: Bruce Richardson 
> CC: Konstantin Ananyev 
> Signed-off-by: Jerin Jacob 
> 
> diff --git a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> index 00b1cdf..4eac666 100644
> --- a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> +++ b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> @@ -61,6 +61,12 @@ extern "C" {
>  
>  #define rte_smp_rmb() rte_compiler_barrier()
>  
> +#define rte_io_mb() rte_mb()
> +
> +#define rte_io_wmb() rte_compiler_barrier()
> +
> +#define rte_io_rmb() rte_compiler_barrier()
> +
>  /*- 16 bit atomic operations
>  * -*/
>  
>  #ifndef RTE_FORCE_INTRINSICS
> 
<...>

[dpdk-dev] [PATCH] config: build nfp pmd support by default

[Alejandro Lucero]

Because using a NFP PMD requires specific BSP installed, the PMD
support was not the default option before. This was just for making
people aware of such dependency, since there is no need for such a
BSP for just compiling DPDK with NFP PMD support.

Signed-off-by: Alejandro Lucero 
---
 config/common_base  |  2 +-
 doc/guides/nics/nfp.rst | 10 ++
 2 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/config/common_base b/config/common_base
index 8e9dcfa..3fd62d6 100644
--- a/config/common_base
+++ b/config/common_base
@@ -248,7 +248,7 @@ CONFIG_RTE_LIBRTE_ENIC_DEBUG=n
 #
 # Compile burst-oriented Netronome NFP PMD driver
 #
-CONFIG_RTE_LIBRTE_NFP_PMD=n
+CONFIG_RTE_LIBRTE_NFP_PMD=y
 CONFIG_RTE_LIBRTE_NFP_DEBUG=n
 
 #
diff --git a/doc/guides/nics/nfp.rst b/doc/guides/nics/nfp.rst
index 4ef6e02..b643e5b 100644
--- a/doc/guides/nics/nfp.rst
+++ b/doc/guides/nics/nfp.rst
@@ -68,10 +68,12 @@ 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:
+Although NFP PMD has Netronome´s BSP dependencies, it is possible to
+compile it along with other DPDK PMDs even if no BSP was installed before.
+Of course, a DPDK app will require such a BSP installed for using the
+NFP PMD.
+
+Default PMD configuration is at **common_linuxapp configuration** file:
 
 - **CONFIG_RTE_LIBRTE_NFP_PMD=y**
 
-- 
1.9.1

[dpdk-dev] [PATCH v4 2/8] net/virtio_user: fix not properly reset device

[Jianfeng Tan]

virtio_user is not properly reset when users call vtpci_reset(),
as it ignores VIRTIO_CONFIG_STATUS_RESET status in
virtio_user_set_status().

This might lead to initialization failure as it starts to re-init
the device before sending RESET messege to backend. Besides, previous
callfds and kickfds are not closed.

To fix it, we add support to disable virtqueues when it's set to
DRIVER OK status, and re-init fields in struct virtio_user_dev.

Fixes: e9efa4d93821 ("net/virtio-user: add new virtual PCI driver")
Fixes: 37a7eb2ae816 ("net/virtio-user: add device emulation layer")

CC: sta...@dpdk.org

Signed-off-by: Jianfeng Tan 
---
 drivers/net/virtio/virtio_user/virtio_user_dev.c | 26 
 drivers/net/virtio/virtio_user_ethdev.c  | 15 --
 2 files changed, 27 insertions(+), 14 deletions(-)

diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c 
b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index 0d7e17b..a38398b 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -182,7 +182,17 @@ virtio_user_start_device(struct virtio_user_dev *dev)
 
 int virtio_user_stop_device(struct virtio_user_dev *dev)
 {
-   return vhost_user_sock(dev->vhostfd, VHOST_USER_RESET_OWNER, NULL);
+   uint32_t i;
+
+   for (i = 0; i < dev->max_queue_pairs * 2; ++i) {
+   close(dev->callfds[i]);
+   close(dev->kickfds[i]);
+   }
+
+   for (i = 0; i < dev->max_queue_pairs; ++i)
+   vhost_user_enable_queue_pair(dev->vhostfd, i, 0);
+
+   return 0;
 }
 
 static inline void
@@ -210,6 +220,8 @@ int
 virtio_user_dev_init(struct virtio_user_dev *dev, char *path, int queues,
 int cq, int queue_size, const char *mac)
 {
+   uint32_t i;
+
snprintf(dev->path, PATH_MAX, "%s", path);
dev->max_queue_pairs = queues;
dev->queue_pairs = 1; /* mq disabled by default */
@@ -218,6 +230,11 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char 
*path, int queues,
parse_mac(dev, mac);
dev->vhostfd = -1;
 
+   for (i = 0; i < VIRTIO_MAX_VIRTQUEUES * 2 + 1; ++i) {
+   dev->kickfds[i] = -1;
+   dev->callfds[i] = -1;
+   }
+
dev->vhostfd = vhost_user_setup(dev->path);
if (dev->vhostfd < 0) {
PMD_INIT_LOG(ERR, "backend set up fails");
@@ -264,13 +281,6 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char 
*path, int queues,
 void
 virtio_user_dev_uninit(struct virtio_user_dev *dev)
 {
-   uint32_t i;
-
-   for (i = 0; i < dev->max_queue_pairs * 2; ++i) {
-   close(dev->callfds[i]);
-   close(dev->kickfds[i]);
-   }
-
close(dev->vhostfd);
 }
 
diff --git a/drivers/net/virtio/virtio_user_ethdev.c 
b/drivers/net/virtio/virtio_user_ethdev.c
index 6c16d2d..298232a 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -87,21 +87,24 @@ virtio_user_write_dev_config(struct virtio_hw *hw, size_t 
offset,
 }
 
 static void
-virtio_user_set_status(struct virtio_hw *hw, uint8_t status)
+virtio_user_reset(struct virtio_hw *hw)
 {
struct virtio_user_dev *dev = virtio_user_get_dev(hw);
 
-   if (status & VIRTIO_CONFIG_STATUS_DRIVER_OK)
-   virtio_user_start_device(dev);
-   dev->status = status;
+   if (dev->status & VIRTIO_CONFIG_STATUS_DRIVER_OK)
+   virtio_user_stop_device(dev);
 }
 
 static void
-virtio_user_reset(struct virtio_hw *hw)
+virtio_user_set_status(struct virtio_hw *hw, uint8_t status)
 {
struct virtio_user_dev *dev = virtio_user_get_dev(hw);
 
-   virtio_user_stop_device(dev);
+   if (status & VIRTIO_CONFIG_STATUS_DRIVER_OK)
+   virtio_user_start_device(dev);
+   else if (status == VIRTIO_CONFIG_STATUS_RESET)
+   virtio_user_reset(hw);
+   dev->status = status;
 }
 
 static uint8_t
-- 
2.7.4

[dpdk-dev] [PATCH v4 3/8] net/virtio_user: move vhost user specific code

[Jianfeng Tan]

To support vhost kernel as the backend of net_virtio_user in coming
patches, we move vhost_user specific structs and macros into
vhost_user.c, and only keep common definitions in vhost.h.

Besides, remove VHOST_USER_MQ feature check.

Signed-off-by: Jianfeng Tan 
---
 drivers/net/virtio/virtio_user/vhost.h   | 36 
 drivers/net/virtio/virtio_user/vhost_user.c  | 32 +
 drivers/net/virtio/virtio_user/virtio_user_dev.c |  9 --
 3 files changed, 32 insertions(+), 45 deletions(-)

diff --git a/drivers/net/virtio/virtio_user/vhost.h 
b/drivers/net/virtio/virtio_user/vhost.h
index 7adb55f..e54ac35 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -42,8 +42,6 @@
 #include "../virtio_logs.h"
 #include "../virtqueue.h"
 
-#define VHOST_MEMORY_MAX_NREGIONS 8
-
 struct vhost_vring_state {
unsigned int index;
unsigned int num;
@@ -105,40 +103,6 @@ struct vhost_memory_region {
uint64_t mmap_offset;
 };
 
-struct vhost_memory {
-   uint32_t nregions;
-   uint32_t padding;
-   struct vhost_memory_region regions[VHOST_MEMORY_MAX_NREGIONS];
-};
-
-struct vhost_user_msg {
-   enum vhost_user_request request;
-
-#define VHOST_USER_VERSION_MASK 0x3
-#define VHOST_USER_REPLY_MASK   (0x1 << 2)
-   uint32_t flags;
-   uint32_t size; /* the following payload size */
-   union {
-#define VHOST_USER_VRING_IDX_MASK   0xff
-#define VHOST_USER_VRING_NOFD_MASK  (0x1 << 8)
-   uint64_t u64;
-   struct vhost_vring_state state;
-   struct vhost_vring_addr addr;
-   struct vhost_memory memory;
-   } payload;
-   int fds[VHOST_MEMORY_MAX_NREGIONS];
-} __attribute((packed));
-
-#define VHOST_USER_HDR_SIZE offsetof(struct vhost_user_msg, payload.u64)
-#define VHOST_USER_PAYLOAD_SIZE \
-   (sizeof(struct vhost_user_msg) - VHOST_USER_HDR_SIZE)
-
-/* The version of the protocol we support */
-#define VHOST_USER_VERSION0x1
-
-#define VHOST_USER_F_PROTOCOL_FEATURES 30
-#define VHOST_USER_MQ (1ULL << VHOST_USER_F_PROTOCOL_FEATURES)
-
 int vhost_user_sock(int vhostfd, enum vhost_user_request req, void *arg);
 int vhost_user_setup(const char *path);
 int vhost_user_enable_queue_pair(int vhostfd, uint16_t pair_idx, int enable);
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c 
b/drivers/net/virtio/virtio_user/vhost_user.c
index 082e821..295ce16 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -42,6 +42,38 @@
 
 #include "vhost.h"
 
+/* The version of the protocol we support */
+#define VHOST_USER_VERSION0x1
+
+#define VHOST_MEMORY_MAX_NREGIONS 8
+struct vhost_memory {
+   uint32_t nregions;
+   uint32_t padding;
+   struct vhost_memory_region regions[VHOST_MEMORY_MAX_NREGIONS];
+};
+
+struct vhost_user_msg {
+   enum vhost_user_request request;
+
+#define VHOST_USER_VERSION_MASK 0x3
+#define VHOST_USER_REPLY_MASK   (0x1 << 2)
+   uint32_t flags;
+   uint32_t size; /* the following payload size */
+   union {
+#define VHOST_USER_VRING_IDX_MASK   0xff
+#define VHOST_USER_VRING_NOFD_MASK  (0x1 << 8)
+   uint64_t u64;
+   struct vhost_vring_state state;
+   struct vhost_vring_addr addr;
+   struct vhost_memory memory;
+   } payload;
+   int fds[VHOST_MEMORY_MAX_NREGIONS];
+} __attribute((packed));
+
+#define VHOST_USER_HDR_SIZE offsetof(struct vhost_user_msg, payload.u64)
+#define VHOST_USER_PAYLOAD_SIZE \
+   (sizeof(struct vhost_user_msg) - VHOST_USER_HDR_SIZE)
+
 static int
 vhost_user_write(int fd, void *buf, int len, int *fds, int fd_num)
 {
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c 
b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index a38398b..8dd563a 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -151,8 +151,6 @@ virtio_user_start_device(struct virtio_user_dev *dev)
 * VIRTIO_NET_F_MAC and VIRTIO_NET_F_CTRL_VQ is stripped.
 */
features = dev->features;
-   if (dev->max_queue_pairs > 1)
-   features |= VHOST_USER_MQ;
features &= ~(1ull << VIRTIO_NET_F_MAC);
features &= ~(1ull << VIRTIO_NET_F_CTRL_VQ);
ret = vhost_user_sock(dev->vhostfd, VHOST_USER_SET_FEATURES, &features);
@@ -268,13 +266,6 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char 
*path, int queues,
dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_MAC_ADDR);
}
 
-   if (dev->max_queue_pairs > 1) {
-   if (!(dev->features & VHOST_USER_MQ)) {
-   PMD_INIT_LOG(ERR, "MQ not supported by the backend");
-   return -1;
-   }
-   }
-
return 0;
 }
 
-- 
2.7.4

[dpdk-dev] [PATCH v4 0/8] virtio_user as an alternative exception path

[Jianfeng Tan]

v4:
  - Fix a clang compiling error by removing "NULL" line in the definition
of vhost_msg_strings. This error does not show up when it's defined
as a static variable, so not necessary to fix it in stable branch.
  - Query kernel to get how many regions are supported, default 64 regions.
  - Set TUNSETSNDBUF to INT_MAX.
  - When get_features, unmask those backend-specific feature bits.
  - Remove VHOST_KERNEL_MAX_QUEUES (8) restriction, but due to another
restriction by VIRTIO_MAX_VIRTQUEUES (8), we still cannot configure
more than 8 queues.
  - Add a howto document.

v3:
  - Drop the patch to postpone driver ok sending patch, superseded it
with a bug fix to disable all virtqueues and re-init the device.
(you might wonder why not just send reset owner msg. Under my test,
 it causes spinlock deadlock problem when killing the program).
  - Avoid compiling error on 32-bit system for pointer convert.
  - Fix a bug in patch "abstract virtio user backend ops", vhostfd is
not properly assigned.
  - Fix a "MQ cannot be used" bug in v2, which is related to strip
some feature bits that vhost kernel does not recognize.
  - Update release note.

v2: (Lots of them are from yuanhan's comment)
  - Add offloding feature.
  - Add multiqueue support.
  - Add a new patch to postpone the sending of driver ok notification.
  - Put fix patch ahead of the whole patch series.
  - Split original 0001 patch into 0003 and 0004 patches.
  - Remove the original vhost_internal design, just add those into
struct virtio_user_dev for simplicity.
  - Reword "control" to "send_request".
  - Reword "host_features" to "device_features". 

In v16.07, we upstreamed a virtual device, virtio_user (with vhost-user
as the backend). The path to go with a vhost-kernel backend has been
dropped for bad performance comparing to vhost-user and code simplicity.

But after a second thought, virtio_user + vhost-kernel is a good 
candidate as an exceptional path, such as KNI, which exchanges packets
with kernel networking stack.
  - maintenance: vhost-net (kernel) is upstreamed and extensively used 
kernel module. We don't need any out-of-tree module like KNI.
  - performance: as with KNI, this solution would use one or more
kthreads to send/receive packets from user space DPDK applications,
which has little impact on user space polling thread (except that
it might enter into kernel space to wake up those kthreads if
necessary).
  - features: vhost-net is born to be a networking solution, which has
lots of networking related featuers, like multi queue, tso, multi-seg
mbuf, etc.


How to test:

Here is a simple test case for:
___testpmd___
iperf-c/iperf-s -- tap0 -- [virtio_user0 -- ixgbe]
  |
  |
  |
 iperf-s/iperf-c -- ixgbe (kernel driver, eth3)

Step 1: Remove MAC addres setting in testpmd csum fwd engine.

diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
index 57e6ae2..225c7a8 100644
--- a/app/test-pmd/csumonly.c
+++ b/app/test-pmd/csumonly.c
@@ -706,10 +706,12 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
 * and inner headers */
 
eth_hdr = rte_pktmbuf_mtod(m, struct ether_hdr *);
+#if 0
ether_addr_copy(&peer_eth_addrs[fs->peer_addr],
ð_hdr->d_addr);
ether_addr_copy(&ports[fs->tx_port].eth_addr,
ð_hdr->s_addr);
+#endif
parse_ethernet(eth_hdr, &info);
l3_hdr = (char *)eth_hdr + info.l2_len;
 
Step 2: bind one ixgbe with igb_uio, and start testpmd
$(testpmd) -c 0xc -n 4 \
--vdev=virtio_user0,path=/dev/vhost-net,queue_size=1024 \
-- -i --txqflags=0x0 --disable-hw-vlan --enable-lro \
--enable-rx-cksum --rxd=1024 --txd=1024

To test multiqueue, start testpmd like this:
$(testpmd) -c 0xc -n 4 \
--vdev=virtio_user0,path=/dev/vhost-net,queues=2,queue_size=1024 \
-- -i --txqflags=0x0 --disable-hw-vlan --enable-lro --enable-rx-cksum \
--txq=2 --rxq=2 --rxd=1024 --txd=1024

Step 3: start testpmd: (port 0 is ixgbe, port 1 is virtio_user)
# set fwd csum
# csum set ip hw 0
# csum set tcp hw 0
# csum set ip sw 1
# csum set tcp hw 1
# tso set 1448 0
# tso set 1448 1
# start

Step 4: start the other end with below script
$ ip netns add ns1
$ ip link set eth3 netns ns1
$ ip netns exec ns1 ifconfig eth3 1.1.1.2/24 up
$ ip netns exec ns1 taskset 0xf0 iperf3 -s -i 1

Use below command if you change the position of iperf-c and iperf-s. 
$ ip netns exec ns1 numactl -N 0 iperf3 -c 1.1.1.3 -i 1 -t 30

Step 5: up the tap0 and configure ip
$ ifconfig 

[dpdk-dev] [PATCH v4 1/8] net/virtio_user: fix wrongly get/set features

[Jianfeng Tan]

Before the commit 86d59b21468a ("net/virtio: support LRO"), features
in virtio PMD, is decided and properly set at device initialization
and will not be changed. But afterward, features could be changed in
virtio_dev_configure(), and will be re-negotiated if it's changed.

In virtio_user, device features is obtained at driver probe phase
only once, but we did not store it. So the added feature bits in
re-negotiation will fail.

To fix it, we store it down, and will be used to feature negotiation
either at device initialization phase or device configure phase.

Fixes: e9efa4d93821 ("net/virtio-user: add new virtual PCI driver")

CC: sta...@dpdk.org

Signed-off-by: Jianfeng Tan 
---
 drivers/net/virtio/virtio_ethdev.h   |  5 
 drivers/net/virtio/virtio_user/virtio_user_dev.c | 34 +++-
 drivers/net/virtio/virtio_user/virtio_user_dev.h |  5 +++-
 drivers/net/virtio/virtio_user_ethdev.c  |  5 ++--
 4 files changed, 28 insertions(+), 21 deletions(-)

diff --git a/drivers/net/virtio/virtio_ethdev.h 
b/drivers/net/virtio/virtio_ethdev.h
index 27d9a19..4feccf9 100644
--- a/drivers/net/virtio/virtio_ethdev.h
+++ b/drivers/net/virtio/virtio_ethdev.h
@@ -70,6 +70,11 @@
 1ULL << VIRTIO_F_VERSION_1   | \
 1ULL << VIRTIO_F_IOMMU_PLATFORM)
 
+#define VIRTIO_PMD_SUPPORTED_GUEST_FEATURES\
+   (VIRTIO_PMD_DEFAULT_GUEST_FEATURES |\
+1u << VIRTIO_NET_F_GUEST_CSUM |\
+1u << VIRTIO_NET_F_GUEST_TSO4 |\
+1u << VIRTIO_NET_F_GUEST_TSO6)
 /*
  * CQ function prototype
  */
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c 
b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index e239e0e..0d7e17b 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -148,12 +148,13 @@ virtio_user_start_device(struct virtio_user_dev *dev)
 
/* Step 1: set features
 * Make sure VHOST_USER_F_PROTOCOL_FEATURES is added if mq is enabled,
-* and VIRTIO_NET_F_MAC is stripped.
+* VIRTIO_NET_F_MAC and VIRTIO_NET_F_CTRL_VQ is stripped.
 */
features = dev->features;
if (dev->max_queue_pairs > 1)
features |= VHOST_USER_MQ;
features &= ~(1ull << VIRTIO_NET_F_MAC);
+   features &= ~(1ull << VIRTIO_NET_F_CTRL_VQ);
ret = vhost_user_sock(dev->vhostfd, VHOST_USER_SET_FEATURES, &features);
if (ret < 0)
goto error;
@@ -228,29 +229,26 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char 
*path, int queues,
}
 
if (vhost_user_sock(dev->vhostfd, VHOST_USER_GET_FEATURES,
-   &dev->features) < 0) {
+   &dev->device_features) < 0) {
PMD_INIT_LOG(ERR, "get_features failed: %s", strerror(errno));
return -1;
}
if (dev->mac_specified)
-   dev->features |= (1ull << VIRTIO_NET_F_MAC);
+   dev->device_features |= (1ull << VIRTIO_NET_F_MAC);
 
-   if (!cq) {
-   dev->features &= ~(1ull << VIRTIO_NET_F_CTRL_VQ);
-   /* Also disable features depends on VIRTIO_NET_F_CTRL_VQ */
-   dev->features &= ~(1ull << VIRTIO_NET_F_CTRL_RX);
-   dev->features &= ~(1ull << VIRTIO_NET_F_CTRL_VLAN);
-   dev->features &= ~(1ull << VIRTIO_NET_F_GUEST_ANNOUNCE);
-   dev->features &= ~(1ull << VIRTIO_NET_F_MQ);
-   dev->features &= ~(1ull << VIRTIO_NET_F_CTRL_MAC_ADDR);
-   } else {
-   /* vhost user backend does not need to know ctrl-q, so
-* actually we need add this bit into features. However,
-* DPDK vhost-user does send features with this bit, so we
-* check it instead of OR it for now.
+   if (cq) {
+   /* device does not really need to know anything about CQ,
+* so if necessary, we just claim to support CQ
 */
-   if (!(dev->features & (1ull << VIRTIO_NET_F_CTRL_VQ)))
-   PMD_INIT_LOG(INFO, "vhost does not support ctrl-q");
+   dev->device_features |= (1ull << VIRTIO_NET_F_CTRL_VQ);
+   } else {
+   dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_VQ);
+   /* Also disable features depends on VIRTIO_NET_F_CTRL_VQ */
+   dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_RX);
+   dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_VLAN);
+   dev->device_features &= ~(1ull << VIRTIO_NET_F_GUEST_ANNOUNCE);
+   dev->device_features &= ~(1ull << VIRTIO_NET_F_MQ);
+   dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_MAC_ADDR);
}
 
if (dev->max_queue_pairs > 1) {
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.h 
b/drivers/net/virtio/virtio_user/virtio_user_dev.h
index 33690b5..28fc788 100644
--- a/dri

Re: [dpdk-dev] [PATCH 1/2] net/virtio: access interrupt handler directly

[Yuanhan Liu]

On Thu, Jan 12, 2017 at 02:03:19PM +0800, Yuanhan Liu wrote:
> Since commit 0e1b45a284b4 ("ethdev: decouple interrupt handling from
> PCI device"), intr_handle is stored at eth_dev struct, that we could
> use it directly. Thus there is no need to get it from hw.
> 
> Signed-off-by: Yuanhan Liu 
> ---

Both applied to dpdk-next-virtio.

--yliu

Re: [dpdk-dev] [PATCH] net/nfp: fix Vlan offload flags check

[Olivier Matz]

On Fri, 13 Jan 2017 12:03:01 +, Ferruh Yigit
 wrote:
> On 1/12/2017 1:04 PM, Olivier Matz wrote:
> > Fix typo when checking that no Vlan offload flags are passed at port
> > initialization.
> > 
> > By the way, also fix a typo in the log.
> > 
> > Fixes: d4a27a3b092a ("nfp: add basic features")
> > 
> > Signed-off-by: Olivier Matz 
> > ---
> >  drivers/net/nfp/nfp_net.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
> > index 65ba09f..ed3c9b8 100644
> > --- a/drivers/net/nfp/nfp_net.c
> > +++ b/drivers/net/nfp/nfp_net.c
> > @@ -2174,8 +2174,8 @@ nfp_net_vlan_offload_set(struct rte_eth_dev
> > *dev, int mask) new_ctrl = 0;
> >  
> > if ((mask & ETH_VLAN_FILTER_OFFLOAD) ||
> > -   (mask & ETH_VLAN_FILTER_OFFLOAD))
> > -   RTE_LOG(INFO, PMD, "Not support for
> > ETH_VLAN_FILTER_OFFLOAD or"
> > +   (mask & ETH_VLAN_FILTER_EXTEND))  
> 
> I can't find "ETH_VLAN_FILTER_EXTEND",
> do you mean "ETH_VLAN_EXTEND_OFFLOAD" ?

Erf... I'm sorry, my mistake, I copy/pasted the log below without
verifying. I saw it while watching the code.

Yes I mean ETH_VLAN_EXTEND_OFFLOAD. By the way, it could even be
replaced ETH_VLAN_*_MASK, but it's the same value. For reference, here
are the definitions from rte_ethdev.h:

 /* Definitions used for VLAN Offload functionality */
 #define ETH_VLAN_STRIP_OFFLOAD   0x0001 /**< VLAN Strip  On/Off */
 #define ETH_VLAN_FILTER_OFFLOAD  0x0002 /**< VLAN Filter On/Off */
 #define ETH_VLAN_EXTEND_OFFLOAD  0x0004 /**< VLAN Extend On/Off */

 /* Definitions used for mask VLAN setting */
 #define ETH_VLAN_STRIP_MASK   0x0001 /**< VLAN Strip  setting mask */
 #define ETH_VLAN_FILTER_MASK  0x0002 /**< VLAN Filter  setting mask*/
 #define ETH_VLAN_EXTEND_MASK  0x0004 /**< VLAN Extend  setting mask*/


I'll submit another patch with ETH_VLAN_EXTEND_OFFLOAD.
Thanks Ferruh.

Olivier

Re: [dpdk-dev] [PATCH] tools: add tags and cscope index file generation support

[Mcnamara, John]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Jerin Jacob
> Sent: Friday, January 13, 2017 2:50 AM
> To: Yigit, Ferruh 
> Cc: dev@dpdk.org; thomas.monja...@6wind.com
> Subject: Re: [dpdk-dev] [PATCH] tools: add tags and cscope index file
> generation support
> 
> On Thu, Jan 12, 2017 at 02:19:44PM +, Ferruh Yigit wrote:
> > On 11/27/2016 12:12 AM, Jerin Jacob wrote:
> > > This script generates cscope, gtags, and tags index files based on
> > > EAL environment.
> > > (architecture and OS(linux/bsd))
> > >
> > > Selection of the architecture and OS environment is based on dpdk
> > > configuration target(T=)
> > >
> > > example usage:
> > > make tags T=x86_64-native-linuxapp-gcc make cscope
> > > T=x86_64-native-linuxapp-gcc make gtags T=x86_64-native-linuxapp-gcc
> > >
> > > Signed-off-by: Jerin Jacob 
> >
> > <...>
> >
> > > +
> > > +if [ -n "$T" ]; then
> > <...>
> > > + usage $1
> > > + exit
> >
> > I would prefer fallback to RTE_TARGET parsing, or processing all files
> > ( cscope -Rqn ), instead of exiting here, but that is your call.
> 
> I would prefer avoid any fallback(sometimes RTE_TARGET can be "build"
> also) and print the usage and exit error if the requirements are not meet.
> 
> >
> >
> > Also "tags.sh" needs to be moved into "devtools" after latest updates.
> 
> Sure. I will send v2 to address this.

Very handy. +1.

Could you also add an etags target, for Emacs. Either separately like the 
gtags target or just add it to the tags target:
 
doetags()
{
all_sources | xargs etags -a
}

# or:

doctags()
{
all_sources | xargs ctags -a
all_sources | xargs etags -a
}

It create a TAGS file which should be added to the gitignore rules.

Thanks,

John

[dpdk-dev] [PATCH v6] app/testpmd: supported offload capabilities query

[Qiming Yang]

Add two new commands "show port cap " and "show
port cap all"to diaplay what offload capabilities supported
in ports. It will not only display all the capabilities of
the port, but also the enabling condition for each capability
in the running time.

Signed-off-by: Qiming Yang 
Acked-by: Jingjing Wu 
Acked-by: Beilei Xing 
---
v2 changes:
* fixed the output style as Ferruh's patch show and add some
  description in docs for new functions.
v3 changes:
* add new command in cmd_help_long_parsed.
v4 changes:
* use 'cap' instead of 'capa'.
v5 changes:
* rebased, fixed the inappropriate expression and adjusted the
  output order.
v6 changes:
* fixed format problems.
---
 app/test-pmd/cmdline.c  |  17 ++-
 app/test-pmd/config.c   | 175 
 app/test-pmd/testpmd.h  |   1 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  12 +-
 4 files changed, 195 insertions(+), 10 deletions(-)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 4e8b0d8..5938daa 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -183,7 +183,7 @@ static void cmd_help_long_parsed(void *parsed_result,
"Display:\n"
"\n\n"
 
-   "show port (info|stats|xstats|fdir|stat_qmap|dcb_tc) 
(port_id|all)\n"
+   "show port 
(info|stats|xstats|fdir|stat_qmap|dcb_tc|cap) (port_id|all)\n"
"Display information for port_id, or all.\n\n"
 
"show port X rss reta (size) (mask0,mask1,...)\n"
@@ -5812,6 +5812,9 @@ static void cmd_showportall_parsed(void *parsed_result,
else if (!strcmp(res->what, "dcb_tc"))
FOREACH_PORT(i, ports)
port_dcb_info_display(i);
+   else if (!strcmp(res->what, "cap"))
+   FOREACH_PORT(i, ports)
+   port_offload_cap_display(i);
 }
 
 cmdline_parse_token_string_t cmd_showportall_show =
@@ -5821,13 +5824,14 @@ cmdline_parse_token_string_t cmd_showportall_port =
TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, port, "port");
 cmdline_parse_token_string_t cmd_showportall_what =
TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, what,
-"info#stats#xstats#fdir#stat_qmap#dcb_tc");
+"info#stats#xstats#fdir#stat_qmap#dcb_tc#cap");
 cmdline_parse_token_string_t cmd_showportall_all =
TOKEN_STRING_INITIALIZER(struct cmd_showportall_result, all, "all");
 cmdline_parse_inst_t cmd_showportall = {
.f = cmd_showportall_parsed,
.data = NULL,
-   .help_str = "show|clear port info|stats|xstats|fdir|stat_qmap|dcb_tc 
all",
+   .help_str = "show|clear port "
+   "info|stats|xstats|fdir|stat_qmap|dcb_tc|cap all",
.tokens = {
(void *)&cmd_showportall_show,
(void *)&cmd_showportall_port,
@@ -5867,6 +5871,8 @@ static void cmd_showport_parsed(void *parsed_result,
nic_stats_mapping_display(res->portnum);
else if (!strcmp(res->what, "dcb_tc"))
port_dcb_info_display(res->portnum);
+   else if (!strcmp(res->what, "cap"))
+   port_offload_cap_display(res->portnum);
 }
 
 cmdline_parse_token_string_t cmd_showport_show =
@@ -5876,14 +5882,15 @@ cmdline_parse_token_string_t cmd_showport_port =
TOKEN_STRING_INITIALIZER(struct cmd_showport_result, port, "port");
 cmdline_parse_token_string_t cmd_showport_what =
TOKEN_STRING_INITIALIZER(struct cmd_showport_result, what,
-"info#stats#xstats#fdir#stat_qmap#dcb_tc");
+"info#stats#xstats#fdir#stat_qmap#dcb_tc#cap");
 cmdline_parse_token_num_t cmd_showport_portnum =
TOKEN_NUM_INITIALIZER(struct cmd_showport_result, portnum, UINT8);
 
 cmdline_parse_inst_t cmd_showport = {
.f = cmd_showport_parsed,
.data = NULL,
-   .help_str = "show|clear port info|stats|xstats|fdir|stat_qmap|dcb_tc "
+   .help_str = "show|clear port "
+   "info|stats|xstats|fdir|stat_qmap|dcb_tc|cap "
"",
.tokens = {
(void *)&cmd_showport_show,
diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c
index 617e6d4..8a2ef56 100644
--- a/app/test-pmd/config.c
+++ b/app/test-pmd/config.c
@@ -542,6 +542,181 @@ port_infos_display(portid_t port_id)
dev_info.tx_desc_lim.nb_min);
printf("TXDs number alignment: %hu\n", dev_info.tx_desc_lim.nb_align);
 }
+void
+port_offload_cap_display(portid_t port_id)
+{
+   struct rte_eth_dev *dev;
+   struct rte_eth_dev_info dev_info;
+   static const char *info_border = "";
+
+   if (port_id_is_invalid(port_id, ENABLED_WARN))
+   return;
+
+   dev = &rte_eth_devices[port_id];
+   rte_eth_dev_info_get(port_id, &dev_info);
+
+   printf(

Re: [dpdk-dev] i40e_aq_get_phy_capabilities() fails when using SFP+ with no link

[Olivier Matz]

Hi,

On Thu, 12 Jan 2017 14:55:54 +0100, Olivier MATZ
 wrote:
> Hi,
> 
> On Wed, 11 Jan 2017 20:51:58 +, "Rowden, Aaron F"
>  wrote:
> > Hi Helin,
> > 
> > I'm checking on this to see why it could be failing but I don’t
> > think this is one part of formal validation. Intel modules are
> > always what is recommended.
> > 
> > Aaron
> >   
> > > Hi Helin, 
> > >   
> > > > On 11 Jan 2017, at 09:08, Zhang, Helin 
> > > > wrote:
> > > > 
> > > > Hi Aaron
> > > > 
> > > > Is the SFP+ (Finisar FTLX8571D3BCL) supported and validated by
> > > > Intel? It seems there is some PHY issue in this case.
> > > 
> > > As the original reporter of this issue, I will test with validated
> > > SFP+s and will report on my testing. 
> > > 
> > > Shouldn’t unsupported SFP+s be blacklisted in the I40E driver? 
> > >   
> 
> Just to let you know that in my case the SFP are Intel ones.
> Maybe it's a different issue.
> 
> I see there are some i40e fixes in the net-next repo, I'll give a try
> with this version.

The issue still exists in net-next.

I did a git bissect, and the commit that introduces the issue is:
f4668a33efe5 ("net/i40e: fix link status change interrupt")  [1]

If I revert it (with some conflicts), the problem I described
in [2] disappear.

Helin, Jinging, do you know what would be the consequences of reverting
this patch?

Christos, I don't know if it also helps for yor issue. If no, sorry for
having squatted your topic, the symptoms looked quite similar at first
glance.

Thanks,
Olivier


[1] http://dpdk.org/browse/dpdk/commit/?id=f4668a33efe5
[2] http://dpdk.org/ml/archives/dev/2017-January/054401.html

Re: [dpdk-dev] [PATCH v5] app/testpmd: supported offload capabilities query

[De Lara Guarch, Pablo]

Hi Qiming,

> -Original Message-
> From: Yang, Qiming
> Sent: Friday, January 13, 2017 9:40 AM
> To: De Lara Guarch, Pablo; dev@dpdk.org
> Subject: RE: [dpdk-dev] [PATCH v5] app/testpmd: supported offload
> capabilities query
> 
> Hi, Pablo
> 
> -Original Message-
> From: De Lara Guarch, Pablo
> Sent: Friday, January 13, 2017 4:43 PM
> To: Yang, Qiming ; dev@dpdk.org
> Cc: Yang, Qiming 
> Subject: RE: [dpdk-dev] [PATCH v5] app/testpmd: supported offload
> capabilities query
> 
> Hi Qiming,
> 
> > -Original Message-
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Qiming Yang
> > Sent: Thursday, January 12, 2017 3:26 AM
> > To: dev@dpdk.org
> > Cc: Yang, Qiming
> > Subject: [dpdk-dev] [PATCH v5] app/testpmd: supported offload
> > capabilities query
> >
> > Add two new commands "show port cap " and "show port cap
> all"to
> > diaplay what offload capabilities supported in ports. It will not only
> > display all the capabilities of the port, but also the enabling
> > condition for each capability in the running time.
> >
> > Signed-off-by: Qiming Yang 
> > Acked-by: Jingjing Wu 
> > Acked-by: Beilei Xing 
> > ---
> > v2 changes:
> > * fixed the output style as Ferruh's patch show and add some
> >   description in docs for new functions.
> > v3 changes:
> > * add new command in cmd_help_long_parsed.
> > v4 changes:
> > * use 'cap' instead of 'capa'.
> > v5 changes:
> > * rebased, fixed the inappropriate expression and adjusted the
> >   output order.
> > ---
> > ---
> >  app/test-pmd/cmdline.c  |  17 ++-
> >  app/test-pmd/config.c   | 175
> > 
> >  app/test-pmd/testpmd.h  |   1 +
> >  doc/guides/testpmd_app_ug/testpmd_funcs.rst |  12 +-
> >  4 files changed, 195 insertions(+), 10 deletions(-)
> >
> > diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c index
> > 4e8b0d8..6fa1783 100644
> > --- a/app/test-pmd/cmdline.c
> > +++ b/app/test-pmd/cmdline.c
> 
> ...
> 
> > +
> > +   if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_CKSUM) {
> > +   printf("RX TCP checksum:   ");
> > +   if (dev->data->dev_conf.rxmode.hw_ip_checksum)
> > +   printf("on\n");
> > +   else
> > +   printf("off\n");
> > +   }
> > +
> > +   if (dev_info.rx_offload_capa &
> > DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM)
> > +   printf("RX Outer IPv4 checksum:");
> 
> Missing on/off?
> Qiming: I didn't find any switch for this feature in DPDK now. I'll fix the
> format problem. Thank you.

I just saw your v6. So, what does this mean? If the device is capable of doing 
this,
but there is no way to enable/disable it, it would mean that it will be enabled 
or disabled by default, right?

I see that X550 NIC is the only one capable of this. Maybe Wenzhuo can help on 
clarifying this.

I would say that, if it is enabled when it is capable, then you should print 
"on",
But printing only "RX Outer IPv4 checksum" does not give any information (when 
all the rest of the capabilities show on/off).

Thanks,
Pablo
 

Re: [dpdk-dev] 17.05 Roadmap

[Bruce Richardson]

On Fri, Jan 13, 2017 at 10:24:01AM +, O'Driscoll, Tim wrote:
> Below are the features that we're planning to submit for the 17.05 release. 
> We'll submit a patch to update the roadmap page with this info.
> 
> It would be good if others are also willing to share their plans so that we 
> can build up a complete picture of what's planned for 17.05 and make sure 
> there's no duplication.
> 
> 
> Cryptodev DES SW PMD: A PMD will be added which implements an optimized 
> software version of the DES algorithm.
> 
> Cryptodev DOCSIS BPI+: The cryptodev API will be enhanced to enable 
> processing of packets according to the Baseline Privacy Interface Plus (BPI+) 
> specification described in the Data-over-Cable Service Interface 
> Specification (DOCSIS) Security Specification 
> (http://www.cablelabs.com/wp-content/uploads/specdocs/CM-SP-SECv3.1-I06-160602.pdf).
>  See the RFC (http://dpdk.org/ml/archives/dev/2016-December/052433.html) for 
> further details. Support will be added to the QAT PMD for AES and DES, to the 
> new software PMD for DES (see previous item), and to the existing AESNI_MB 
> PMD.
> 
> Cryptodev Scheduler: This allows packets to be encrypted/decrypted in either 
> SW or HW depending on packet size and HW utilization. Reordering will be 
> supported so that packet order can be preserved.
> 
> Ethernet 32-bit CRC Generation: An optimized x86 library for CRC-32 will be 
> added. A CRC may need to be generated for a downstream frame by the DOCSIS 
> MAC layer because the CRC may be removed by the NIC or the frame may be 
> modified by the MAC for PHS (packet header suppression). 
> 
> API to Configure Programmable Devices: More devices are now supporting 
> programmable components, for example the Pipeline Personalization Profiles in 
> I40E. An API will be added to allow any programmable device to be configured.
> 
> I40E Hardware QoS: Hardware QoS will be supported on the I40E. This will 
> include Tx bandwidth control (min and max), and Rx Traffic Class assignment.
> 
> Configurable Tunnel Filters for I40E: DPDK support will be added for a new 
> I40E admin queue which allows configuration of filter types for cloud filters.
> 
> Enable MPLS on I40E: MPLSoUDP and MPLSoGRE will be supported for I40E, 
> including the new protocols and filtering (RSS and Flow Director), checksum 
> offload and packet type recognition.
> 
> Software Eventdev Implementation: The libeventdev API is being added in 17.02 
> (http://dpdk.org/ml/archives/dev/2016-December/052877.html). A software 
> implementation of this API will be added.

Just a minor clarification based on my understanding and following some
discussion with Thomas.

The eventdev API is being added to the next-eventdev tree, not the main
tree for 17.02, as it is not useful without an implementation to back it
up. As well as that, once implementations are produced by multiple
vendors we may see the APIs evolve slightly.
Therefore, it's only in 17.05 that we are going to see the eventdev API
in the mainline, along with at least one - but hopefully more -
implementations backing it up.

Regards,
/Bruce

Re: [dpdk-dev] [RFC PATCH 00/11] generalise rte_ring to allow different datatypes

[Olivier Matz]

Hi Bruce,

On Wed, 11 Jan 2017 15:05:14 +, Bruce Richardson
 wrote:
> The rte_ring library in DPDK provides an excellent high-performance
> mechanism which can be used for passing pointers between cores and
> for other tasks such as buffering. However, it does have a number
> of limitations:
> 
> * type information of pointers is lost, as it works with void pointers
> * typecasting is needed when using enqueue/dequeue burst functions,
>   since arrays of other types cannot be automatically cast to void **
> * the data to be passed through the ring itself must be no bigger than
>   a pointer
> 
> While the first two limitations are an inconvenience, the final one is
> one that can prevent use of rte_rings in cases where their
> functionality is needed. The use-case which has inspired the patchset
> is that of eventdev. When working with rte_events, each event is a
> 16-byte structure consisting of a pointer and some metadata e.g.
> priority and type. For these events, what is passed around between
> cores is not pointers to events, but the events themselves. This
> makes existing rings unsuitable for use by applications working with
> rte_events, and also for use internally inside any software
> implementation of an eventdev.
> 
> For rings to handle events or other similarly sized structures, e.g.
> NIC descriptors, etc., we then have two options - duplicate rte_ring
> code to create new ring implementations for each of those types, or
> generalise the existing code using macros so that the data type
> handled by each rings is a compile time paramter. This patchset takes
> the latter approach, and once applied would allow us to add an
> rte_event_ring type to DPDK using a header file containing:
> 
> #define RING_TYPE struct rte_event
> #define RING_TYPE_NAME rte_event
> #include 
> #undef RING_TYPE_NAME
> #undef RING_TYPE
> 
> [NOTE: the event_ring is not defined in this set, since it depends on
> the eventdev implementation not present in the main tree]
> 
> If we want to elimiate some of the typecasting on our code when
> enqueuing and dequeuing mbuf pointers, an rte_mbuf_ring type can be
> similarly created using the same number of lines of code.
> 
> The downside of this generalisation is that the code for the rings now
> has far more use of macros in it. However, I do not feel that overall
> readability suffers much from this change, the since the changes are
> pretty much just search-replace onces. There should also be no ABI
> compatibility issues with this change, since the existing rte_ring
> structures remain the same.

I didn't dive deeply in the patches, just had a quick look. I
understand the need, and even if I really don't like the "#define +
#include" way to create a new specific ring (for readability,
grepability), that may be a solution to your problem.

I think using a similar approach than in sys/queue.h would be even
worse in terms of readability.


What do you think about the following approach?

- add a new elt_size in rte_ring structure

- update create/enqueue/dequeue/... functions to manage the elt size

- change:
rte_ring_enqueue_bulk(struct rte_ring *r,
  void * const *obj_table, unsigned n)
  to:
rte_ring_enqueue_bulk(struct rte_ring *r, void *obj_table,
  unsigned n) 

  This relaxes the type for the API in the function. In the caller,
  the type of obj_table would be:
  - (void **) in case of a ring of pointers
  - (uint8_t *) in case of a ring of uint8_t
  - (struct rte_event *) in case of a ring of rte_event
  ...

  I think (I have not tested it) it won't break compilation since
  any type can be implicitly casted into a void *. Also, I'd say it
  is possible to avoid breaking the ABI.

- deprecate or forbid calls to:
rte_ring_mp_enqueue(struct rte_ring *r, void *obj)
(and similar)

  Because with a ring of pointers, obj is the pointer, passed by value.
  For other types, we would need
rte_ring_mp_enqueue(struct rte_ring *r,  obj)

  Maybe we could consider using a macro here.


The drawbacks I see are:
- a dynamic elt_size may slightly decrease performance
- it still uses casts to (void *), so there is no type checking


Regards,
Olivier

[dpdk-dev] [PATCH] scripts: skip capitalization check for commit prefixes

[Bruce Richardson]

The prefix in the commit title must be a valid component name and is
checked in separate checks. For capitalization, just check the part after
the colon. This is already done for most capitalization checks, just make
the remainder consistent with this.

Signed-off-by: Bruce Richardson 
---
 scripts/check-git-log.sh | 8 
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/scripts/check-git-log.sh b/scripts/check-git-log.sh
index f79f0a2..f6a35d2 100755
--- a/scripts/check-git-log.sh
+++ b/scripts/check-git-log.sh
@@ -113,10 +113,10 @@ bad=$(echo "$headlines" | grep --color=always \
 
 # check headline uppercase (Rx/Tx, VF, L2, MAC, Linux, ARM...)
 bad=$(echo "$headlines" | grep -E --color=always \
-   -e '\<(rx|tx|RX|TX)\>' \
-   -e '\<[pv]f\>' \
-   -e '\<[hsf]w\>' \
-   -e '\' \
+   -e ':.*\<(rx|tx|RX|TX)\>' \
+   -e ':.*\<[pv]f\>' \
+   -e ':.*\<[hsf]w\>' \
+   -e ':.*\' \
-e ':.*\' \
-e ':.*\' \
-e ':.*\' \
-- 
2.9.3

Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device memory read/write API

[Ananyev, Konstantin]

> -Original Message-
> From: Richardson, Bruce
> Sent: Friday, January 13, 2017 11:47 AM
> To: Yigit, Ferruh 
> Cc: Jerin Jacob ; dev@dpdk.org; Ananyev, 
> Konstantin ;
> thomas.monja...@6wind.com; jianbo@linaro.org; vikto...@rehivetech.com; 
> santosh.shu...@caviumnetworks.com; Griffin, John
> ; Trahe, Fiona ; Jain, Deepak 
> K 
> Subject: Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device 
> memory read/write API
> 
> On Fri, Jan 13, 2017 at 11:40:06AM +, Ferruh Yigit wrote:
> > On 1/13/2017 8:17 AM, Jerin Jacob wrote:
> > > On Thu, Jan 12, 2017 at 07:09:22PM +, Ferruh Yigit wrote:
> > >> Hi Jerin,
> > >>
> > >> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> > >> <...>
> > >>
> > >>> +#include 
> > >>> +
> > >>>  /* CSR write macro */
> > >>> -#define ADF_CSR_WR(csrAddr, csrOffset, val) \
> > >>> -   (void)((*((volatile uint32_t *)(((uint8_t *)csrAddr) + 
> > >>> csrOffset)) \
> > >>> -   = (val)))
> > >>> +#define ADF_CSR_WR(csrAddr, csrOffset, val)\
> > >>> +   rte_write32(val, (((uint8_t *)csrAddr) + csrOffset))
> > >>
> > >> For IA, this update introduces an extra compiler barrier (rte_io_wmb()),
> > >> which is indeed not a must, is this correct?
> > >
> > > AFAIK, Compiler barrier is required for IA. I am not an IA expert, if
> > > someone thinks it needs to changed then I can fix it in following commit
> > > in this patch series by making rte_io_wmb() and rte_io_rmb() as empty.
> > >
> > > Let me know.
> >
> > I don't know, but what I know is this was working for IA without
> > compiler barrier before.
> >
> > Bruce or Konstantin can help here.
> >
> I think having a compiler barrier is safer. If all data being written
> before the actual register write to the device is volatile, none is
> needed, but also in that case, the compiler barrier should have no
> effect. 

+1

>If some of the data is not volatile, then a barrier is needed
> for correctness. IMHO, unless there is a performance regression by doing
> so, I think having the IO register writes have compiler barriers on IA
> is a good thing. It should save individual drivers from having to worry
> about the barriers themselves in most cases.
> 
> Regards,
> /Bruce

Re: [dpdk-dev] [PATCH] net/nfp: fix Vlan offload flags check

[Ferruh Yigit]

On 1/12/2017 1:04 PM, Olivier Matz wrote:
> Fix typo when checking that no Vlan offload flags are passed at port
> initialization.
> 
> By the way, also fix a typo in the log.
> 
> Fixes: d4a27a3b092a ("nfp: add basic features")
> 
> Signed-off-by: Olivier Matz 
> ---
>  drivers/net/nfp/nfp_net.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
> index 65ba09f..ed3c9b8 100644
> --- a/drivers/net/nfp/nfp_net.c
> +++ b/drivers/net/nfp/nfp_net.c
> @@ -2174,8 +2174,8 @@ nfp_net_vlan_offload_set(struct rte_eth_dev *dev, int 
> mask)
>   new_ctrl = 0;
>  
>   if ((mask & ETH_VLAN_FILTER_OFFLOAD) ||
> - (mask & ETH_VLAN_FILTER_OFFLOAD))
> - RTE_LOG(INFO, PMD, "Not support for ETH_VLAN_FILTER_OFFLOAD or"
> + (mask & ETH_VLAN_FILTER_EXTEND))

I can't find "ETH_VLAN_FILTER_EXTEND",
do you mean "ETH_VLAN_EXTEND_OFFLOAD" ?

> + RTE_LOG(INFO, PMD, "No support for ETH_VLAN_FILTER_OFFLOAD or"
>   " ETH_VLAN_FILTER_EXTEND");
>  
>   /* Enable vlan strip if it is not configured yet */
> 

[dpdk-dev] [PATCH v1] doc: add guidelines on stable and lts releases

[John McNamara]

Add document explaining the current Stable and LTS process.

Signed-off-by: John McNamara 
---

V1: For background see previous discussions on Stable and LTS releses:

   http://dpdk.org/ml/archives/dev/2016-July/044848.html
   http://dpdk.org/ml/archives/dev/2016-June/040256.html

doc/guides/contributing/index.rst  |  1 +
 doc/guides/contributing/stable.rst | 99 ++
 2 files changed, 100 insertions(+)
 create mode 100644 doc/guides/contributing/stable.rst

diff --git a/doc/guides/contributing/index.rst 
b/doc/guides/contributing/index.rst
index f6af317..329b678 100644
--- a/doc/guides/contributing/index.rst
+++ b/doc/guides/contributing/index.rst
@@ -10,4 +10,5 @@ Contributor's Guidelines
 versioning
 documentation
 patches
+stable
 cheatsheet
diff --git a/doc/guides/contributing/stable.rst 
b/doc/guides/contributing/stable.rst
new file mode 100644
index 000..735e116
--- /dev/null
+++ b/doc/guides/contributing/stable.rst
@@ -0,0 +1,99 @@
+.. stable_lts_releases:
+
+DPDK Stable Releases and Long Term Support
+==
+
+This section sets out the guidelines for the DPDK Stable Releases and the DPDK
+Long Term Support releases (LTS).
+
+
+Introduction
+
+
+The purpose of the DPDK Stable Releases is to maintain releases of DPDK with
+backported fixes over an extended period of time. This provides downstream
+consumers of DPDK with a stable target on which to base applications or
+packages.
+
+The Long Term Support release (LTS) is a designation applied to a Stable
+Release to indicate longer term support.
+
+
+Stable Releases
+---
+
+Any major release of DPDK can be designated as a Stable Release if a
+maintainer volunteers to maintain it.
+
+A Stable Release is used to backport fixes from an ``N`` release back to an
+``N-1`` release, for example, from 16.11 to 16.07.
+
+The duration of a stable is one complete release cycle (3 months). It can be
+longer, up to 1 year, if a maintainer continues to support the stable branch,
+or if users supply backported fixes, however the explicit commitment should be
+for one release cycle.
+
+The release cadence is determined by the maintainer based on the number of
+bugfixes and the criticality of the bugs. Releases should be coordinated with
+the validation engineers to ensure that a tagged release has been tested.
+
+
+LTS Release
+---
+
+A stable release can be designated as an LTS release based on community
+agreement and a commitment from a maintainer. An LTS release will have a
+maintenance duration of 2 years.
+
+The current DPDK LTS release is 16.11.
+
+It is anticipated that there will be at least 4 releases per year of the LTS
+or approximately 1 every 3 months. However, the cadence can be shorter or
+longer depending on the number and criticality of the backported
+fixes. Releases should be coordinated with the validation engineers to ensure
+that a tagged release has been tested.
+
+
+What changes should be backported
+-
+
+Backporting should be limited to bug fixes.
+
+Features should not be backported to stable releases. It may be acceptable, in
+limited cases, to back port features for the LTS release where:
+
+* There is a justifiable use case (for example a new PMD).
+* The change is non-invasive.
+* The work of preparing the backport is done by the proposer.
+* There is support within the community.
+
+
+The Stable Mailing List
+---
+
+The Stable and LTS release are coordinated on the sta...@dpdk.org mailing
+list.
+
+All fix patches to the master branch that are candidates for backporting
+should also be CCed to the `sta...@dpdk.org 
`_
+mailing list.
+
+
+Releasing
+-
+
+A Stable Release will be released by:
+
+* Tagging the release with YY.MM.n (year, month, number).
+* Uploading a tarball of the release to dpdk.org.
+* Sending an announcement to the `annou...@dpdk.org 
`_
+  list.
+
+Stable release are available on the `dpdk.org download page 
`_.
+
+
+ABI
+---
+
+The Stable Release should not be seen as a way of breaking or circumventing
+the DPDK ABI policy.
-- 
2.7.4

Re: [dpdk-dev] [PATCH v1] doc: add guidelines on stable and lts releases

[Mcnamara, John]

> -Original Message-
> From: Mcnamara, John
> Sent: Friday, January 13, 2017 1:06 PM
> To: dev@dpdk.org
> Cc: Liu, Yuanhan ; thomas.monja...@6wind.com;
> Mcnamara, John 
> Subject: [PATCH v1] doc: add guidelines on stable and lts releases
> 
> Add document explaining the current Stable and LTS process.

Hi,

Just a reminder that DPDK 16.07 was a stable release, 16.11 is a stable
release and it will also become the 2 year LTS release.

Some links:

http://dpdk.org/ml/archives/stable/ - Mailing list
http://dpdk.org/download - Download page for mainline and stable release
http://fast.dpdk.org/rel/ - Stable release tarballs

If you are a downstream consumer of DPDK and would like to influence how the
Stable/LTS releases are working, or should work, then let us know.

John

[dpdk-dev] [PATCH v4 6/8] net/virtio_user: enable offloading

[Jianfeng Tan]

When used with vhost kernel backend, we can offload at both directions.
  - From vhost kernel to virtio_user, the offload is enabled so that
DPDK app can trust the flow is checksum-correct; and if DPDK app
sends it through another port, the checksum needs to be
recalculated or offloaded. It also applies to TSO.
  - From virtio_user to vhost_kernel, the offload is enabled so that
kernel can trust the flow is L4-checksum-correct, no need to verify
it; if kernel will consume it, DPDK app should make sure the
l3-checksum is correctly set.

Signed-off-by: Jianfeng Tan 
---
 drivers/net/virtio/virtio_user/vhost_kernel.c | 40 +--
 drivers/net/virtio/virtio_user/vhost_kernel_tap.c | 14 
 drivers/net/virtio/virtio_user/vhost_kernel_tap.h |  7 
 3 files changed, 59 insertions(+), 2 deletions(-)

diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c 
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 9301f96..a5d3e1e 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -165,6 +165,28 @@ prepare_vhost_memory_kernel(void)
return vm;
 }
 
+/* with below features, vhost kernel does not need to do the checksum and TSO,
+ * these info will be passed to virtio_user through virtio net header.
+ */
+#define VHOST_KERNEL_GUEST_OFFLOADS_MASK   \
+   ((1ULL << VIRTIO_NET_F_GUEST_CSUM) |\
+(1ULL << VIRTIO_NET_F_GUEST_TSO4) |\
+(1ULL << VIRTIO_NET_F_GUEST_TSO6) |\
+(1ULL << VIRTIO_NET_F_GUEST_ECN)  |\
+(1ULL << VIRTIO_NET_F_GUEST_UFO))
+
+/* with below features, when flows from virtio_user to vhost kernel
+ * (1) if flows goes up through the kernel networking stack, it does not need
+ * to verify checksum, which can save CPU cycles;
+ * (2) if flows goes through a Linux bridge and outside from an interface
+ * (kernel driver), checksum and TSO will be done by GSO in kernel or even
+ * offloaded into real physical device.
+ */
+#define VHOST_KERNEL_HOST_OFFLOADS_MASK\
+   ((1ULL << VIRTIO_NET_F_HOST_TSO4) | \
+(1ULL << VIRTIO_NET_F_HOST_TSO6) | \
+(1ULL << VIRTIO_NET_F_CSUM))
+
 static int
 vhost_kernel_ioctl(struct virtio_user_dev *dev,
   enum vhost_user_request req,
@@ -186,10 +208,15 @@ vhost_kernel_ioctl(struct virtio_user_dev *dev,
arg = (void *)vm;
}
 
-   /* We don't need memory protection here */
-   if (req_kernel == VHOST_SET_FEATURES)
+   if (req_kernel == VHOST_SET_FEATURES) {
+   /* We don't need memory protection here */
*(uint64_t *)arg &= ~(1ULL << VIRTIO_F_IOMMU_PLATFORM);
 
+   /* VHOST kernel does not know about below flags */
+   *(uint64_t *)arg &= ~VHOST_KERNEL_GUEST_OFFLOADS_MASK;
+   *(uint64_t *)arg &= ~VHOST_KERNEL_HOST_OFFLOADS_MASK;
+   }
+
for (i = 0; i < dev->max_queue_pairs; ++i) {
if (dev->vhostfds[i] < 0)
continue;
@@ -199,6 +226,15 @@ vhost_kernel_ioctl(struct virtio_user_dev *dev,
break;
}
 
+   if (!ret && req_kernel == VHOST_GET_FEATURES) {
+   /* with tap as the backend, all these features are supported
+* but not claimed by vhost-net, so we add them back when
+* reporting to upper layer.
+*/
+   *((uint64_t *)arg) |= VHOST_KERNEL_GUEST_OFFLOADS_MASK;
+   *((uint64_t *)arg) |= VHOST_KERNEL_HOST_OFFLOADS_MASK;
+   }
+
if (vm)
free(vm);
 
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel_tap.c 
b/drivers/net/virtio/virtio_user/vhost_kernel_tap.c
index 797713b..cdb5c3c 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel_tap.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel_tap.c
@@ -50,6 +50,12 @@ vhost_kernel_open_tap(char **p_ifname, int hdr_size)
int sndbuf = INT_MAX;
struct ifreq ifr;
int tapfd;
+   unsigned int offload =
+   TUN_F_CSUM |
+   TUN_F_TSO4 |
+   TUN_F_TSO6 |
+   TUN_F_TSO_ECN |
+   TUN_F_UFO;
 
/* TODO:
 * 1. verify we can get/set vnet_hdr_len, tap_probe_vnet_hdr_len
@@ -106,6 +112,14 @@ vhost_kernel_open_tap(char **p_ifname, int hdr_size)
goto error;
}
 
+   /* TODO: before set the offload capabilities, we'd better (1) check
+* negotiated features to see if necessary to offload; (2) query tap
+* to see if it supports the offload capabilities.
+*/
+   if (ioctl(tapfd, TUNSETOFFLOAD, offload) != 0)
+   PMD_DRV_LOG(ERR, "TUNSETOFFLOAD ioctl() failed: %s",
+  strerror(errno));
+
if (!(*p_ifname))
*p_ifname = strdup(ifr.ifr_name);
 
diff --git a/drivers/net/virtio/virtio_

Re: [dpdk-dev] [PATCH] net/nfp: fix Vlan offload flags check

[Ferruh Yigit]

On 1/13/2017 12:48 PM, Olivier Matz wrote:
> On Fri, 13 Jan 2017 12:03:01 +, Ferruh Yigit
>  wrote:
>> On 1/12/2017 1:04 PM, Olivier Matz wrote:
>>> Fix typo when checking that no Vlan offload flags are passed at port
>>> initialization.
>>>
>>> By the way, also fix a typo in the log.
>>>
>>> Fixes: d4a27a3b092a ("nfp: add basic features")
>>>
>>> Signed-off-by: Olivier Matz 
>>> ---

<...>

> 
> 
> I'll submit another patch with ETH_VLAN_EXTEND_OFFLOAD.

If this will be the only change, I can update it if you want:

-   (mask & ETH_VLAN_FILTER_OFFLOAD))
-   RTE_LOG(INFO, PMD, "Not support for ETH_VLAN_FILTER_OFFLOAD or"
-   " ETH_VLAN_FILTER_EXTEND");
+   (mask & ETH_VLAN_EXTEND_OFFLOAD))
+   RTE_LOG(INFO, PMD, "No support for ETH_VLAN_FILTER_OFFLOAD or"
+   " ETH_VLAN_EXTEND_OFFLOAD");

> Thanks Ferruh.
> 
> Olivier
> 

[dpdk-dev] [PATCH v4 4/8] net/virtio_user: abstract virtio user backend ops

[Jianfeng Tan]

Add a struct virtio_user_backend_ops to abstract three kinds of backend
operations:
  - setup, create the unix socket connection;
  - send_request, sync messages with backend;
  - enable_qp, enable some queue pair.

Signed-off-by: Jianfeng Tan 
---
 drivers/net/virtio/virtio_user/vhost.h   | 17 +-
 drivers/net/virtio/virtio_user/vhost_user.c  | 66 +++-
 drivers/net/virtio/virtio_user/virtio_user_dev.c | 77 +++-
 drivers/net/virtio/virtio_user/virtio_user_dev.h |  5 ++
 4 files changed, 106 insertions(+), 59 deletions(-)

diff --git a/drivers/net/virtio/virtio_user/vhost.h 
b/drivers/net/virtio/virtio_user/vhost.h
index e54ac35..515e4fc 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -96,6 +96,8 @@ enum vhost_user_request {
VHOST_USER_MAX
 };
 
+const char * const vhost_msg_strings[VHOST_USER_MAX];
+
 struct vhost_memory_region {
uint64_t guest_phys_addr;
uint64_t memory_size; /* bytes */
@@ -103,8 +105,17 @@ struct vhost_memory_region {
uint64_t mmap_offset;
 };
 
-int vhost_user_sock(int vhostfd, enum vhost_user_request req, void *arg);
-int vhost_user_setup(const char *path);
-int vhost_user_enable_queue_pair(int vhostfd, uint16_t pair_idx, int enable);
+struct virtio_user_dev;
+
+struct virtio_user_backend_ops {
+   int (*setup)(struct virtio_user_dev *dev);
+   int (*send_request)(struct virtio_user_dev *dev,
+   enum vhost_user_request req,
+   void *arg);
+   int (*enable_qp)(struct virtio_user_dev *dev,
+uint16_t pair_idx,
+int enable);
+};
 
+struct virtio_user_backend_ops ops_user;
 #endif
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c 
b/drivers/net/virtio/virtio_user/vhost_user.c
index 295ce16..4ad7b21 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -41,6 +41,7 @@
 #include 
 
 #include "vhost.h"
+#include "virtio_user_dev.h"
 
 /* The version of the protocol we support */
 #define VHOST_USER_VERSION0x1
@@ -255,24 +256,25 @@ prepare_vhost_memory_user(struct vhost_user_msg *msg, int 
fds[])
 
 static struct vhost_user_msg m;
 
-static const char * const vhost_msg_strings[] = {
-   [VHOST_USER_SET_OWNER] = "VHOST_USER_SET_OWNER",
-   [VHOST_USER_RESET_OWNER] = "VHOST_USER_RESET_OWNER",
-   [VHOST_USER_SET_FEATURES] = "VHOST_USER_SET_FEATURES",
-   [VHOST_USER_GET_FEATURES] = "VHOST_USER_GET_FEATURES",
-   [VHOST_USER_SET_VRING_CALL] = "VHOST_USER_SET_VRING_CALL",
-   [VHOST_USER_SET_VRING_NUM] = "VHOST_USER_SET_VRING_NUM",
-   [VHOST_USER_SET_VRING_BASE] = "VHOST_USER_SET_VRING_BASE",
-   [VHOST_USER_GET_VRING_BASE] = "VHOST_USER_GET_VRING_BASE",
-   [VHOST_USER_SET_VRING_ADDR] = "VHOST_USER_SET_VRING_ADDR",
-   [VHOST_USER_SET_VRING_KICK] = "VHOST_USER_SET_VRING_KICK",
-   [VHOST_USER_SET_MEM_TABLE] = "VHOST_USER_SET_MEM_TABLE",
-   [VHOST_USER_SET_VRING_ENABLE] = "VHOST_USER_SET_VRING_ENABLE",
-   NULL,
+const char * const vhost_msg_strings[] = {
+   [VHOST_USER_SET_OWNER] = "VHOST_SET_OWNER",
+   [VHOST_USER_RESET_OWNER] = "VHOST_RESET_OWNER",
+   [VHOST_USER_SET_FEATURES] = "VHOST_SET_FEATURES",
+   [VHOST_USER_GET_FEATURES] = "VHOST_GET_FEATURES",
+   [VHOST_USER_SET_VRING_CALL] = "VHOST_SET_VRING_CALL",
+   [VHOST_USER_SET_VRING_NUM] = "VHOST_SET_VRING_NUM",
+   [VHOST_USER_SET_VRING_BASE] = "VHOST_SET_VRING_BASE",
+   [VHOST_USER_GET_VRING_BASE] = "VHOST_GET_VRING_BASE",
+   [VHOST_USER_SET_VRING_ADDR] = "VHOST_SET_VRING_ADDR",
+   [VHOST_USER_SET_VRING_KICK] = "VHOST_SET_VRING_KICK",
+   [VHOST_USER_SET_MEM_TABLE] = "VHOST_SET_MEM_TABLE",
+   [VHOST_USER_SET_VRING_ENABLE] = "VHOST_SET_VRING_ENABLE",
 };
 
-int
-vhost_user_sock(int vhostfd, enum vhost_user_request req, void *arg)
+static int
+vhost_user_sock(struct virtio_user_dev *dev,
+   enum vhost_user_request req,
+   void *arg)
 {
struct vhost_user_msg msg;
struct vhost_vring_file *file = 0;
@@ -280,9 +282,9 @@ vhost_user_sock(int vhostfd, enum vhost_user_request req, 
void *arg)
int fds[VHOST_MEMORY_MAX_NREGIONS];
int fd_num = 0;
int i, len;
+   int vhostfd = dev->vhostfd;
 
RTE_SET_USED(m);
-   RTE_SET_USED(vhost_msg_strings);
 
PMD_DRV_LOG(INFO, "%s", vhost_msg_strings[req]);
 
@@ -403,15 +405,13 @@ vhost_user_sock(int vhostfd, enum vhost_user_request req, 
void *arg)
 
 /**
  * Set up environment to talk with a vhost user backend.
- * @param path
- *   - The path to vhost user unix socket file.
  *
  * @return
- *   - (-1) if fail to set up;
- *   - (>=0) if successful, and it is the fd to vhostfd.
+ *   - (-1) if fail;
+ *   - (0) if succeed.
  */
-int
-vhost_user_setup(const char *path)
+static int
+vhost_user_setup(struct virtio_

Re: [dpdk-dev] [PATCH] net/nfp: fix Vlan offload flags check

[Olivier Matz]

On Fri, 13 Jan 2017 13:17:17 +, Ferruh Yigit
 wrote:
> On 1/13/2017 12:48 PM, Olivier Matz wrote:
> > On Fri, 13 Jan 2017 12:03:01 +, Ferruh Yigit
> >  wrote:  
> >> On 1/12/2017 1:04 PM, Olivier Matz wrote:  
> >>> Fix typo when checking that no Vlan offload flags are passed at
> >>> port initialization.
> >>>
> >>> By the way, also fix a typo in the log.
> >>>
> >>> Fixes: d4a27a3b092a ("nfp: add basic features")
> >>>
> >>> Signed-off-by: Olivier Matz 
> >>> ---  
> 
> <...>
> 
> > 
> > 
> > I'll submit another patch with ETH_VLAN_EXTEND_OFFLOAD.  
> 
> If this will be the only change, I can update it if you want:
> 
> -   (mask & ETH_VLAN_FILTER_OFFLOAD))
> -   RTE_LOG(INFO, PMD, "Not support for ETH_VLAN_FILTER_OFFLOAD
> or"
> -   " ETH_VLAN_FILTER_EXTEND");
> +   (mask & ETH_VLAN_EXTEND_OFFLOAD))
> +   RTE_LOG(INFO, PMD, "No support for ETH_VLAN_FILTER_OFFLOAD or"
> +   " ETH_VLAN_EXTEND_OFFLOAD");
> 

Looks good to me, thank you.

Olivier

[dpdk-dev] [PATCH v4 7/8] net/virtio_user: enable multiqueue with vhost kernel

[Jianfeng Tan]

With vhost kernel, to enable multiqueue, we need backend device
in kernel support multiqueue feature. Specifically, with tap
as the backend, as linux/Documentation/networking/tuntap.txt shows,
we check if tap supports IFF_MULTI_QUEUE feature.

And for vhost kernel, each queue pair has a vhost fd, and with a tap
fd binding this vhost fd. All tap fds are set with the same tap
interface name.

Signed-off-by: Jianfeng Tan 
---
 drivers/net/virtio/virtio_user/vhost_kernel.c | 68 ---
 drivers/net/virtio/virtio_user/vhost_kernel_tap.c |  5 +-
 drivers/net/virtio/virtio_user/vhost_kernel_tap.h |  2 +-
 drivers/net/virtio/virtio_user/virtio_user_dev.c  |  1 +
 4 files changed, 67 insertions(+), 9 deletions(-)

diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c 
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index a5d3e1e..0e45dd6 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -188,6 +188,29 @@ prepare_vhost_memory_kernel(void)
 (1ULL << VIRTIO_NET_F_CSUM))
 
 static int
+tap_supporte_mq(void)
+{
+   int tapfd;
+   unsigned int tap_features;
+
+   tapfd = open(PATH_NET_TUN, O_RDWR);
+   if (tapfd < 0) {
+   PMD_DRV_LOG(ERR, "fail to open %s: %s",
+   PATH_NET_TUN, strerror(errno));
+   return -1;
+   }
+
+   if (ioctl(tapfd, TUNGETFEATURES, &tap_features) == -1) {
+   PMD_DRV_LOG(ERR, "TUNGETFEATURES failed: %s", strerror(errno));
+   close(tapfd);
+   return -1;
+   }
+
+   close(tapfd);
+   return tap_features & IFF_MULTI_QUEUE;
+}
+
+static int
 vhost_kernel_ioctl(struct virtio_user_dev *dev,
   enum vhost_user_request req,
   void *arg)
@@ -196,6 +219,8 @@ vhost_kernel_ioctl(struct virtio_user_dev *dev,
unsigned int i;
uint64_t req_kernel;
struct vhost_memory_kernel *vm = NULL;
+   int vhostfd;
+   unsigned int queue_sel;
 
PMD_DRV_LOG(INFO, "%s", vhost_msg_strings[req]);
 
@@ -215,15 +240,37 @@ vhost_kernel_ioctl(struct virtio_user_dev *dev,
/* VHOST kernel does not know about below flags */
*(uint64_t *)arg &= ~VHOST_KERNEL_GUEST_OFFLOADS_MASK;
*(uint64_t *)arg &= ~VHOST_KERNEL_HOST_OFFLOADS_MASK;
+
+   *(uint64_t *)arg &= ~(1ULL << VIRTIO_NET_F_MQ);
}
 
-   for (i = 0; i < dev->max_queue_pairs; ++i) {
-   if (dev->vhostfds[i] < 0)
-   continue;
+   switch (req_kernel) {
+   case VHOST_SET_VRING_NUM:
+   case VHOST_SET_VRING_ADDR:
+   case VHOST_SET_VRING_BASE:
+   case VHOST_GET_VRING_BASE:
+   case VHOST_SET_VRING_KICK:
+   case VHOST_SET_VRING_CALL:
+   queue_sel = *(unsigned int *)arg;
+   vhostfd = dev->vhostfds[queue_sel / 2];
+   *(unsigned int *)arg = queue_sel % 2;
+   PMD_DRV_LOG(DEBUG, "vhostfd=%d, index=%u",
+   vhostfd, *(unsigned int *)arg);
+   break;
+   default:
+   vhostfd = -1;
+   }
+   if (vhostfd == -1) {
+   for (i = 0; i < dev->max_queue_pairs; ++i) {
+   if (dev->vhostfds[i] < 0)
+   continue;
 
-   ret = ioctl(dev->vhostfds[i], req_kernel, arg);
-   if (ret < 0)
-   break;
+   ret = ioctl(dev->vhostfds[i], req_kernel, arg);
+   if (ret < 0)
+   break;
+   }
+   } else {
+   ret = ioctl(vhostfd, req_kernel, arg);
}
 
if (!ret && req_kernel == VHOST_GET_FEATURES) {
@@ -233,6 +280,12 @@ vhost_kernel_ioctl(struct virtio_user_dev *dev,
 */
*((uint64_t *)arg) |= VHOST_KERNEL_GUEST_OFFLOADS_MASK;
*((uint64_t *)arg) |= VHOST_KERNEL_HOST_OFFLOADS_MASK;
+
+   /* vhost_kernel will not declare this feature, but it does
+* support multi-queue.
+*/
+   if (tap_supporte_mq())
+   *(uint64_t *)arg |= (1ull << VIRTIO_NET_F_MQ);
}
 
if (vm)
@@ -305,6 +358,7 @@ vhost_kernel_enable_queue_pair(struct virtio_user_dev *dev,
int hdr_size;
int vhostfd;
int tapfd;
+   int req_mq = (dev->max_queue_pairs > 1);
 
vhostfd = dev->vhostfds[pair_idx];
 
@@ -324,7 +378,7 @@ vhost_kernel_enable_queue_pair(struct virtio_user_dev *dev,
else
hdr_size = sizeof(struct virtio_net_hdr);
 
-   tapfd = vhost_kernel_open_tap(&dev->ifname, hdr_size);
+   tapfd = vhost_kernel_open_tap(&dev->ifname, hdr_size, req_mq);
if (tapfd < 0) {
PMD_DRV_LOG(ERR, "fail to open tap for vhost kernel");
return -1;
diff --git a/drivers/net/virtio/virtio_user/vhos

[dpdk-dev] [PATCH v4 8/8] doc: add guide to use virtio_user as exceptional path

[Jianfeng Tan]

Signed-off-by: Jianfeng Tan 
---
 .../img/virtio_user_as_excpetional_path.png| Bin 0 -> 38600 bytes
 doc/guides/prog_guide/index.rst|   1 +
 .../prog_guide/virtio_user_as_exceptional_path.rst | 104 +
 3 files changed, 105 insertions(+)
 create mode 100644 
doc/guides/prog_guide/img/virtio_user_as_excpetional_path.png
 create mode 100644 doc/guides/prog_guide/virtio_user_as_exceptional_path.rst

diff --git a/doc/guides/prog_guide/img/virtio_user_as_excpetional_path.png 
b/doc/guides/prog_guide/img/virtio_user_as_excpetional_path.png
new file mode 100644
index 
..9fd45e13eefc2ef39ac2c389f407d60ab4144c59
GIT binary patch
literal 38600
zcmeFZdpy(qA2&W76`>TJkPbu+9mM6hkRnDo=QL5tVO(L(+mv$(Q4t~M97nNb*gBA;
znuUfnD`#ee88+K~AL_cUuIqcae~0`2J?=m5KN9o#ybrJA^Ywha-xAK7pWU)~_ht|X
zw8g~uv?U0%&Hw~j(=50F_(Yg*vi-TI8vT#u~x5|-Ak&q(l2h%zHLE{W-)EUd$ev#N^M`@`f-z!0zOOqRz4}dL?_;qa&t8o(
z5M2EzW6NgQ)o)zY{}(*;+u^)FCcKo~EUC{rw>lRH^QkS>`=GlNYJ>
zT&CZ=vYC1EeP6>z?}rknnDq`~D{mg%$_^D*-`YW3ev*$(p=2M}%{5hgSA*T}slTpA
zjfMMqVm)GNtm9PztX#qxcphYSI&0xAhthhI9lGOi`>peAiCwOpi{y>%+k3K==#BT!
z($2Y4%48m_xRC4qK#*V5(xRg{Q#FUz#Mkodz_vbC;_!G{O2q8U9Y&QW>>es-e
z>%qd|;BiM7Q~3gfHhW?IWFDJ8qt}JqGZtFZ&Itc_*f~~O+?jWB8&#Ce5387e{w9}E
zJ!O|UajbLhTzYvs*}a(a#Mjw*SVvVoAv;$u%ba+`+nY*RQ$AMjMnrSYdFy)AMwYf-
zgr1$RGa-8D_R~AaM>(<53{jjJqczM0zU0N|vfD@}GM$*CSII%F-u^{6b&RSJc02GP
zo9%|~?z2Vs(EJoWskfrtKQGMJDb+#C`}$5pb0_ED_s>3|SYxZI{Id@L;&MDPFC8i+
zOo_HCh^o_D3z1qJKmF-}-42)`B>^vDJ+GfEB6fS}7LFo$|g7w2$ia
zZj;A3XORzWNbvEY@^aaaiICL32auA{+jVsx4(g6@PIuD_%(b}Qo_i5@yeJyAgkHM6
zGgHTVE{rjKjWlP)(JuqHwhU?^u;kBZ?uGHXClCT@6s~N1xGCtG{((B`^;$-sv^yA!
zQ75tLG?BZhabae-cZeilt3Jv{HrSK)NeesXU2@hs`s~WZz?2ekCr-J&j$n6!kCc=#
z#eE*_!m>MgM}>&IiW=@>QKsv3*>=Y`*eXX#9=LOi&sme8Qc@|4u~}kO9qGU7jGih9
z>2Nx^32Sge&u(J}GMCe^J5ZOEZp>-YMYl(vOCQcflQ3S|fl^v2$?{IP2_%U0!_?cKX38hZ}n84n<%$?aOGLs-4MQPeHG`4X?F)rm+CaYG|Yr%u{(
zvU2=*n!9!Vm?9e>%N%v?hg?qB#yHc6Q$7yv8K+pvlnP&VJNrdrllQy{g1_O?m8!
z*VOfTbS@1-;mGb9@;0vDEopB@yo8jFuHU2%nOvoK{gMOO=45C$>^q{Ql+ZFt(8Ciw
z#OA3keB6}K_P)3Y_LVPwecSz*@Ymq8oTn-!xA0Xx*E`5y-3_cSI7BgrCK<=%!)ew{
z8^(z93%m^Pyuyzuqd@L@Vf#too{UZ}-edOKpO*O)$k=9b#+h8Oyv|a7hfS;=uyI%kvA8qf0;P1cU8}-k
z?6hu6k~_88iC9#Y@FYhFxsA#?*-Pl}p2sel-@zH2+JnGqAf3t0>+dV#(32rv6a8}H
z({_y4koAxvF1czG{`>hl8ep#>cYMRM<_|L#c=a}ZX}j2afk@hOj)!)Tkor0=B}RYz
z;Xz51yHrJ8QOm9@SL-cLxCd=16t$OKFKcyOI)gZY=RJv~nJFSomG5}|rEACGi`?#@
zeN*GSx>4_1!B22a9y-M)3J(5_Y;uYWUrb0Tp1}AMa^M)F@tIGfwmZ(BdnB3FjOq61
z>|JM>Y+;AD7|Rw@2ofeUusq*S0?1FCq6Dc?Yv^qM?ZgKi7#7bYo`N-^icB
zeO>SdM#X!U;?x%#HaAqA9ruT8GndQ4xI2Gt<@$|NRtt7a%%p0RmaC7VW#QPVzP_}~
zLbeIK6ZZNY?J4yNCot8USdQxs8H05TRC8VAH`BkT%~nNS8D5AIc~7pt-)CW&$-7xK
z?jILRN{G^hTGdrza^P5_lRGR{baSV!+aqmUuP&%+E_0M~+U63GrrR;UlRa#=!(n2K
zT|uycJ`j4j+;o_8}?pG5!1cb&y-A56;1k)>Nd@_0@272;BQr
z;OxW){AJE|KxsKd8=rV`vRHuNGVJ<~e9$Ci);=X;HAY-H1_
zj5bh%ybZHnS?y?Q&`-}3FV4t)T+yI
zew^NG-;y2GyR3CU4>@88q0cbg+fN8o1ud!NZ2
z0#b!LY`*t0{<|&w>{%(4iXG(1{N4q04O^?LSd?Bv4mg|JIqJKnAwWNU#~g)L@_f9q
zq3Mkand9^9DcwFIJ&Mp^&Kbik`v-=**MGJuY5uX+ic_xZ*K>t%TdJ4@=c$~dJk?pv
zZK9vx)i8p$HpnUX#h9O;Io0_n6Qjt-FX^pcQzY(1kpytg?^&;6A(I^5a1Vcnt_vFC
z(kFo}Q`<5L(=FX(rn0i$G_M7@xnR{oc91wixn`8S!)`v9SCd&hIG(t@e%k_8J-D{Q
zJ)2>}gCnT>8%bmZ=cNAo-^S^=l?BsOS9mUED3PqWP-!=?>W&WT3am5(uzi<_JDD=e
zsIT9~ulr2<)IXxE!&9X@%irUv-0Ip#LZ%-nC74kt=}h~f#~h+X7rY)41<$f)B4p}!
zD5%ZwYIB2?ZCfg?6;LwV=q3#R%KS&DpI|L8}JD=igChChc`43Q>3S
zM)CNR42ND~p2Pl*{Sc0XvYI$A`?GQtpA0!!+_xNUDV_=5eNp`)MO({q*--|!eJb$g
zbYe9br_q~k4=?sPzKHuUU7re!eoZP5?q
z-w*uizP7Xr_6qs9D5_U|?OCvXhsjql9Ivi&yedh{R>`=@xu6-N$Z4Mw%bh?6@9tS=
z%g@UEo#XhK({FTX{r%@p3h32A%-(d|UXJ(g-TwX+CFZ4bZk_f5?>;Rkn;Xu_TG-78_rxPAOlehvkZHWVOoS)5_>|1-A&^$GdF_
zIeJFfT8BOYRbgahVbtk8#x_)iZyT$%u0LUW=Xi)OsGt-W1{*(pE>>sEU?YRC_-!i9
zR1yZ_Z(sj{?`=i=`FojNdDx<4S6v6
zqv}UpJYKm6AGL-^#P-`lqj$~*?Dx^OY9)(4hi`lKF=!&dC$*y8m>TNVT1u>B+YWVD
zLqvRZ>At3W409#Nh`C3clkVm;giSGZ<=7`LBF~KNRa)~1#~a=mn2!@
zswG^BdZLcHLB8>bdaklG>&rOk;{)oI>e8s=9MdrBECXEw9^DJi7Kis=(C;k
zCYd%?a{A4+LSM_F)!t-mi
zE^?qQxO}r0?_&av9E*Ct9cfL%Msrbx2L=shpzvvFsh!T0TJY+3@xAbMMh#cb>
zOzxIeNL_1iQ~L`)L+a&APhd89!5WbUQ-fs3?wL=UZdV?dPj|@5pda@M4VzM9@I(Cb
ziJqCliv(*rzCye8o-~w()COnuZx|*Qcv=aQOwfeHQ)HWkMpX}V@(rB3+E1t*SUt4S
z0=7grHST&&{+>->J+rQT>A`nBbP6RD>ixhNuWYk?_?W#
zP+kub)ecl7z5B49ZqP%>hIkcXU(pQeTZLkL@48d`@<~Hch9*u#S2Zw{4dr50$QI_z
zdYGKRJHnka1?`{QzbzD1on5G>rQ@7Ge^y69k{0gxCekz{vJa0|8vGDOgM%80e8jm}8b#A!(rg9>G9<(+2;KPC+p)8n!AWxws{d=;mvevQL~pH0TP
zoM0|sq-hIqM4uif*jd)Pg-(KfLilms`a%9v07(=q7IYFPYfXw1eb0+
z7TE`{wlUiH04et69*z=u9~#^V-F&}_CJnW#H6h(7)w9_605;Tbdu6R%c1UHrWMEl7
z#op#g^`?Y6eac>o@AQ054ZYB1;|xo3sCew)sg3M+ePwhXL$bJqUS_^gBF0GSWOL|^
zh=Es*R?I~WDH&ZF;$^iWu0qb4lf}v^-k%MbnkQ&7sV>lzrx-dtoPsZ>|&#{Y<#
zg5ZyPt^$zys-s_)e!jT@V+Mi+x7I?~jhHmA`vH#kC1h&1ApD@81sYV2bj!gkZ1M~i
z9_p&}i)zh0aVQq4ou7f-vazX!;BJ>8O6i|X5^CBqipN#!av;{s(9-#9%;)3KEcz`Z
z+&@?Ht!c7%cfAj(p%j-x)OrYkjLdsDLxbn*%qSTSIcHEMqlhfBk+SjFQVu665{|Va
zBEl=C-TO-B>(E|Vbzr!{&cYsVcP}ZnRx$UpOfYK8gg=CIWlVEux<9_4y^CC1Mjn*Y
zT#03%%f)LA8V)`8zV7Cw4s(R(!x)|?##zB4dqn
zA^1qw#^N_gcbbgkT9=C3s|T~)w@q-qbZ{?8Ib-dFPO!AB=0o^>MZ0cJa||U9alIpX
zDBmpPN>~ByG3AF{6Ter5

[dpdk-dev] [PATCH v4 5/8] net/virtio_user: add vhost kernel support

[Jianfeng Tan]

This patch add support vhost kernel as the backend for virtio_user.
Three main hook functions are added:
  - vhost_kernel_setup() to open char device, each vq pair needs one
vhostfd;
  - vhost_kernel_ioctl() to communicate control messages with vhost
kernel module;
  - vhost_kernel_enable_queue_pair() to open tap device and set it
as the backend of corresonding vhost fd (that is to say, vq pair).

Signed-off-by: Jianfeng Tan 
---
 doc/guides/rel_notes/release_17_02.rst|  20 ++
 drivers/net/virtio/Makefile   |   2 +
 drivers/net/virtio/virtio_user/vhost.h|   2 +
 drivers/net/virtio/virtio_user/vhost_kernel.c | 311 ++
 drivers/net/virtio/virtio_user/vhost_kernel_tap.c | 116 
 drivers/net/virtio/virtio_user/vhost_kernel_tap.h |  60 +
 drivers/net/virtio/virtio_user/virtio_user_dev.c  |  36 ++-
 drivers/net/virtio/virtio_user/virtio_user_dev.h  |   5 +
 8 files changed, 549 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/virtio/virtio_user/vhost_kernel.c
 create mode 100644 drivers/net/virtio/virtio_user/vhost_kernel_tap.c
 create mode 100644 drivers/net/virtio/virtio_user/vhost_kernel_tap.h

diff --git a/doc/guides/rel_notes/release_17_02.rst 
b/doc/guides/rel_notes/release_17_02.rst
index 180af82..7354df5 100644
--- a/doc/guides/rel_notes/release_17_02.rst
+++ b/doc/guides/rel_notes/release_17_02.rst
@@ -52,6 +52,26 @@ New Features
   See the :ref:`Generic flow API ` documentation for more
   information.
 
+* **virtio_user with vhost-kernel as another exceptional path.**
+
+  Previously, we upstreamed a virtual device, virtio_user with vhost-user
+  as the backend, as a way for IPC (Inter-Process Communication) and user
+  space container networking.
+
+  Virtio_user with vhost-kernel as the backend is a solution for exceptional
+  path, such as KNI, which exchanges packets with kernel networking stack.
+  This solution is very promising in:
+
+  * maintenance: vhost and vhost-net (kernel) is upstreamed and extensively
+used kernel module.
+  * features: vhost-net is born to be a networking solution, which has
+lots of networking related featuers, like multi queue, tso, multi-seg
+mbuf, etc.
+  * performance: similar to KNI, this solution would use one or more
+kthreads to send/receive packets from user space DPDK applications,
+which has little impact on user space polling thread (except that
+it might enter into kernel space to wake up those kthreads if
+necessary).
 
 Resolved Issues
 ---
diff --git a/drivers/net/virtio/Makefile b/drivers/net/virtio/Makefile
index 97972a6..8b5b5d6 100644
--- a/drivers/net/virtio/Makefile
+++ b/drivers/net/virtio/Makefile
@@ -60,6 +60,8 @@ endif
 
 ifeq ($(CONFIG_RTE_VIRTIO_USER),y)
 SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user/vhost_user.c
+SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user/vhost_kernel.c
+SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user/vhost_kernel_tap.c
 SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user/virtio_user_dev.c
 SRCS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio_user_ethdev.c
 endif
diff --git a/drivers/net/virtio/virtio_user/vhost.h 
b/drivers/net/virtio/virtio_user/vhost.h
index 515e4fc..5c983bd 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -118,4 +118,6 @@ struct virtio_user_backend_ops {
 };
 
 struct virtio_user_backend_ops ops_user;
+struct virtio_user_backend_ops ops_kernel;
+
 #endif
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c 
b/drivers/net/virtio/virtio_user/vhost_kernel.c
new file mode 100644
index 000..9301f96
--- /dev/null
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -0,0 +1,311 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2016 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 

Re: [dpdk-dev] [RFC PATCH 00/11] generalise rte_ring to allow different datatypes

[Bruce Richardson]

On Fri, Jan 13, 2017 at 03:23:34PM +0100, Olivier Matz wrote:
> Hi Bruce,
> 
> On Wed, 11 Jan 2017 15:05:14 +, Bruce Richardson
>  wrote:
> > The rte_ring library in DPDK provides an excellent high-performance
> > mechanism which can be used for passing pointers between cores and
> > for other tasks such as buffering. However, it does have a number
> > of limitations:
> > 
> > * type information of pointers is lost, as it works with void pointers
> > * typecasting is needed when using enqueue/dequeue burst functions,
> >   since arrays of other types cannot be automatically cast to void **
> > * the data to be passed through the ring itself must be no bigger than
> >   a pointer
> > 
> > While the first two limitations are an inconvenience, the final one is
> > one that can prevent use of rte_rings in cases where their
> > functionality is needed. The use-case which has inspired the patchset
> > is that of eventdev. When working with rte_events, each event is a
> > 16-byte structure consisting of a pointer and some metadata e.g.
> > priority and type. For these events, what is passed around between
> > cores is not pointers to events, but the events themselves. This
> > makes existing rings unsuitable for use by applications working with
> > rte_events, and also for use internally inside any software
> > implementation of an eventdev.
> > 
> > For rings to handle events or other similarly sized structures, e.g.
> > NIC descriptors, etc., we then have two options - duplicate rte_ring
> > code to create new ring implementations for each of those types, or
> > generalise the existing code using macros so that the data type
> > handled by each rings is a compile time paramter. This patchset takes
> > the latter approach, and once applied would allow us to add an
> > rte_event_ring type to DPDK using a header file containing:
> > 
> > #define RING_TYPE struct rte_event
> > #define RING_TYPE_NAME rte_event
> > #include 
> > #undef RING_TYPE_NAME
> > #undef RING_TYPE
> > 
> > [NOTE: the event_ring is not defined in this set, since it depends on
> > the eventdev implementation not present in the main tree]
> > 
> > If we want to elimiate some of the typecasting on our code when
> > enqueuing and dequeuing mbuf pointers, an rte_mbuf_ring type can be
> > similarly created using the same number of lines of code.
> > 
> > The downside of this generalisation is that the code for the rings now
> > has far more use of macros in it. However, I do not feel that overall
> > readability suffers much from this change, the since the changes are
> > pretty much just search-replace onces. There should also be no ABI
> > compatibility issues with this change, since the existing rte_ring
> > structures remain the same.
> 
> I didn't dive deeply in the patches, just had a quick look. I
> understand the need, and even if I really don't like the "#define +
> #include" way to create a new specific ring (for readability,
> grepability), that may be a solution to your problem.
> 
> I think using a similar approach than in sys/queue.h would be even
> worse in terms of readability.
> 
> 
> What do you think about the following approach?
> 
> - add a new elt_size in rte_ring structure
> 
> - update create/enqueue/dequeue/... functions to manage the elt size
> 
> - change:
> rte_ring_enqueue_bulk(struct rte_ring *r,
>   void * const *obj_table, unsigned n)
>   to:
> rte_ring_enqueue_bulk(struct rte_ring *r, void *obj_table,
>   unsigned n) 
> 
>   This relaxes the type for the API in the function. In the caller,
>   the type of obj_table would be:
>   - (void **) in case of a ring of pointers
>   - (uint8_t *) in case of a ring of uint8_t
>   - (struct rte_event *) in case of a ring of rte_event
>   ...
> 
>   I think (I have not tested it) it won't break compilation since
>   any type can be implicitly casted into a void *. Also, I'd say it
>   is possible to avoid breaking the ABI.
> 
> - deprecate or forbid calls to:
> rte_ring_mp_enqueue(struct rte_ring *r, void *obj)
> (and similar)
> 
>   Because with a ring of pointers, obj is the pointer, passed by value.
>   For other types, we would need
> rte_ring_mp_enqueue(struct rte_ring *r,  obj)
> 
>   Maybe we could consider using a macro here.
> 
> 
> The drawbacks I see are:
> - a dynamic elt_size may slightly decrease performance
> - it still uses casts to (void *), so there is no type checking
>

Hi Olivier,

Thanks for the feedback.

Yes, I thought about that parameterized sizes solution too, but I did
not pursue it primarily because I was worried about the performance
hits. It would mean that the actual copies of the data elements would
have to be done via memcpy calls - or switches based on size - rather
than assignments, as now. Given that all these calls to enqueue/dequeue
are inlined, that could really hurt performance, as the size of the
elements to be copied are unknown to the compiler at compile time - as
the size is stored in the struct, and not

Re: [dpdk-dev] [PATCH] net/nfp: fix Vlan offload flags check

[Alejandro Lucero]

I was assuming in my acked answer, that the patch had followed usual path
and automatic compilation did not fail.

But I forgot about the NFP PMD not being enabled by default :-)

Hopefully, this will change if this just sent patch is accepted:

http://www.dpdk.org/ml/archives/dev/2017-January/054890.html

By other hand, I have compiled the NFP PMD with this patch and it works.

So acked again.

Thanks


On Fri, Jan 13, 2017 at 2:48 PM, Olivier Matz 
wrote:

> On Fri, 13 Jan 2017 13:17:17 +, Ferruh Yigit
>  wrote:
> > On 1/13/2017 12:48 PM, Olivier Matz wrote:
> > > On Fri, 13 Jan 2017 12:03:01 +, Ferruh Yigit
> > >  wrote:
> > >> On 1/12/2017 1:04 PM, Olivier Matz wrote:
> > >>> Fix typo when checking that no Vlan offload flags are passed at
> > >>> port initialization.
> > >>>
> > >>> By the way, also fix a typo in the log.
> > >>>
> > >>> Fixes: d4a27a3b092a ("nfp: add basic features")
> > >>>
> > >>> Signed-off-by: Olivier Matz 
> > >>> ---
> >
> > <...>
> >
> > >
> > >
> > > I'll submit another patch with ETH_VLAN_EXTEND_OFFLOAD.
> >
> > If this will be the only change, I can update it if you want:
> >
> > -   (mask & ETH_VLAN_FILTER_OFFLOAD))
> > -   RTE_LOG(INFO, PMD, "Not support for ETH_VLAN_FILTER_OFFLOAD
> > or"
> > -   " ETH_VLAN_FILTER_EXTEND");
> > +   (mask & ETH_VLAN_EXTEND_OFFLOAD))
> > +   RTE_LOG(INFO, PMD, "No support for ETH_VLAN_FILTER_OFFLOAD or"
> > +   " ETH_VLAN_EXTEND_OFFLOAD");
> >
>
> Looks good to me, thank you.
>
> Olivier
>

Re: [dpdk-dev] [PATCH v3 15/29] crypto/qat: use eal I/O device memory read/write API

[Jerin Jacob]

On Fri, Jan 13, 2017 at 11:32:29AM +, Ferruh Yigit wrote:
> On 1/13/2017 8:17 AM, Jerin Jacob wrote:
> > On Thu, Jan 12, 2017 at 07:09:22PM +, Ferruh Yigit wrote:
> >> Hi Jerin,
> >>
> >> On 1/12/2017 9:17 AM, Jerin Jacob wrote:
> >> <...>
> >>
> >>> +#include 
> >>> +
> >>>  /* CSR write macro */
> >>> -#define ADF_CSR_WR(csrAddr, csrOffset, val) \
> >>> - (void)((*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)) \
> >>> - = (val)))
> >>> +#define ADF_CSR_WR(csrAddr, csrOffset, val)  \
> >>> + rte_write32(val, (((uint8_t *)csrAddr) + csrOffset))
> >>
> >> For IA, this update introduces an extra compiler barrier (rte_io_wmb()),
> >> which is indeed not a must, is this correct?
> > 
> > AFAIK, Compiler barrier is required for IA. I am not an IA expert, if
> > someone thinks it needs to changed then I can fix it in following commit
> > in this patch series by making rte_io_wmb() and rte_io_rmb() as empty.
> > 
> > Let me know.
> > 
> > AFAIK, Linux kernel code has a barrier in readl/writel for IA.
> > 
> > Typically we don't use any non relaxed versions in fast path.In fast
> > typically all the drivers has explicit write barrier for doorbell write
> > and followed by a relaxed version of write. IMO, In any event, it won't
> > generate performance regression.
> > 
> > [dpdk-master] $ git show
> > 70c343bdc8c33a51a9db23cd58122bdfc120a58f
> > commit 70c343bdc8c33a51a9db23cd58122bdfc120a58f
> > Author: Jerin Jacob 
> > Date:   Mon Dec 5 06:36:49 2016 +0530
> > 
> > eal/x86: define I/O device memory barriers for IA
> > 
> > The patch does not provide any functional change for IA.
> > I/O barriers are mapped to existing smp barriers.
> > 
> > CC: Bruce Richardson 
> > CC: Konstantin Ananyev 
> > Signed-off-by: Jerin Jacob 
> > 
> > diff --git a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > index 00b1cdf..4eac666 100644
> > --- a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > +++ b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > @@ -61,6 +61,12 @@ extern "C" {
> >  
> >  #define rte_smp_rmb() rte_compiler_barrier()
> >  
> > +#define rte_io_mb() rte_mb()
> > +
> > +#define rte_io_wmb() rte_compiler_barrier()
> > +
> > +#define rte_io_rmb() rte_compiler_barrier()
> > +
> >  /*- 16 bit atomic operations
> >  * -*/
> >  
> >  #ifndef RTE_FORCE_INTRINSICS
> > 
> >>
> >> If so, does it make sense to override these functions for x86, and make
> >> rte_writeX = rte_writeX_relaxed
> >> rte_readX = rte_readX_relaxed
> >>
> >>>  
> >>>  /* CSR read macro */
> >>> -#define ADF_CSR_RD(csrAddr, csrOffset) \
> >>> - (*((volatile uint32_t *)(((uint8_t *)csrAddr) + csrOffset)))
> >>> +#define ADF_CSR_RD(csrAddr, csrOffset)   \
> >>> + rte_read32uint8_t *)csrAddr) + csrOffset))
> >>
> >> This patchset both introduces new rte_readX/rte_writeX functions, also
> >> applies them into drivers.
> >>
> >> While applying them, it changes the behavior.
> >> Like above code was doing a read, but after update it does read and
> >> read_memory_barrier.
> >>
> >> What do you think this patchset updates usage in a manner that keeps
> >> behavior exact same. Like using rte_read32_relaxed for this case.
> >> And doing architecture related updates in a different patchset?
> > 
> > Need to use rte_read32 at this commit otherwise it will break for ARM.
> > That's was all point for this patchset.
> 
> Why it breaks the ARM, is it because rte_*mb() updated for ARM in this
> patchset (patch 7/29) ?

Yes.


> 
> I believe it is good to make these modifications in two phase:

It is in two phases only. First introduced the API with implementation and
enabled in each driver. Why did you think other-way around it is better?
I can rework and test if there is any value addition. If you concerned
about git bisect ability then I don't think we are loosing that in this
model.

Thoughts?

> - First replace old usage with rte_readX/rte_writeX while keeping exact
> same behavior
> 
> - Second, do architecture specific changes. Both in eal and drivers
> level if required.
> 
> Thanks,
> ferruh
> 
> > For performance regression, we can always verify by taking delta
> > between this changeset and the previous changeset. If you think, I need
> > to make rte_io_wmb()/rte_io_rmb() as empty for IA then I could do that
> > as well.
> > 
> > 
> >>
> >> This both makes easy to see architecture specific updates, and makes
> >> easy to trace any possible performance issues by this patchset.
> >>
> >>>  
> >>>  #define ADF_BANK_INT_SRC_SEL_MASK_0 0x444CUL
> >>>  #define ADF_BANK_INT_SRC_SEL_MASK_X 0xUL
> >>>
> >>
> 

[dpdk-dev] [PATCH] net/nfp: fix VLAN offload flags check

[Ferruh Yigit]

From: Olivier Matz 

Fix typo when checking that no VLAN offload flags are passed at port
initialization.

By the way, also fix a typo in the log.

Fixes: d4a27a3b092a ("nfp: add basic features")

Signed-off-by: Olivier Matz 
Acked-by: Alejandro Lucero 
---
 drivers/net/nfp/nfp_net.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index 31e72e2..ab2d331 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -2174,9 +2174,9 @@ nfp_net_vlan_offload_set(struct rte_eth_dev *dev, int 
mask)
new_ctrl = 0;
 
if ((mask & ETH_VLAN_FILTER_OFFLOAD) ||
-   (mask & ETH_VLAN_FILTER_OFFLOAD))
-   RTE_LOG(INFO, PMD, "Not support for ETH_VLAN_FILTER_OFFLOAD or"
-   " ETH_VLAN_FILTER_EXTEND");
+   (mask & ETH_VLAN_EXTEND_OFFLOAD))
+   RTE_LOG(INFO, PMD, "No support for ETH_VLAN_FILTER_OFFLOAD or"
+   " ETH_VLAN_EXTEND_OFFLOAD");
 
/* Enable vlan strip if it is not configured yet */
if ((mask & ETH_VLAN_STRIP_OFFLOAD) &&
-- 
2.9.3

Re: [dpdk-dev] [PATCH] net/nfp: fix VLAN offload flags check

[Ferruh Yigit]

On 1/13/2017 3:16 PM, Ferruh Yigit wrote:
> From: Olivier Matz 
> 
> Fix typo when checking that no VLAN offload flags are passed at port
> initialization.
> 
> By the way, also fix a typo in the log.
> 
> Fixes: d4a27a3b092a ("nfp: add basic features")
> 
> Signed-off-by: Olivier Matz 
> Acked-by: Alejandro Lucero 

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

(I forget about stable tree, that is why need to re-send latest patch)

Re: [dpdk-dev] [PATCH v4 1/6] lib: distributor performance enhancements

[Bruce Richardson]

On Mon, Jan 09, 2017 at 07:50:43AM +, David Hunt wrote:
> Now sends bursts of up to 8 mbufs to each worker, and tracks
> the in-flight flow-ids (atomic scheduling)
> 
> New file with a new api, similar to the old API except with _burst
> at the end of the function names
> 
> Signed-off-by: David Hunt 
> ---
>  lib/librte_distributor/Makefile|   2 +
>  lib/librte_distributor/rte_distributor.c   |  72 +--
>  lib/librte_distributor/rte_distributor_burst.c | 558 
> +
>  lib/librte_distributor/rte_distributor_burst.h | 255 ++
>  lib/librte_distributor/rte_distributor_priv.h  | 189 +++
>  lib/librte_distributor/rte_distributor_version.map |   9 +
>  6 files changed, 1014 insertions(+), 71 deletions(-)
>  create mode 100644 lib/librte_distributor/rte_distributor_burst.c
>  create mode 100644 lib/librte_distributor/rte_distributor_burst.h
>  create mode 100644 lib/librte_distributor/rte_distributor_priv.h
> 
Run a documentation sanity check after this patch throws up a few
warnings:

--- /dev/null   2017-01-10 10:26:01.206201474 +
+++ /tmp/doc-check/doc.txt  2017-01-13 15:19:50.717102848 +
@@ -0,0 +1,6 @@
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_burst.h:187:
warning: argument 'mbuf' of command @param is not found in the argument
list of rte_distributor_return_pkt_burst(struct rte_distributor_burst
*d, unsigned int worker_id, struct rte_mbuf **oldpkt, int num)
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_burst.h:199:
warning: The following parameters of
rte_distributor_return_pkt_burst(struct rte_distributor_burst *d,
unsigned int worker_id, struct rte_mbuf **oldpkt, int num) are not
documented:
+  parameter 'oldpkt'
+  parameter 'num'
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_priv.h:73:
warning: Found unknown command `\in_flight_bitmask'
+/home/bruce/dpdk-clean/lib/librte_distributor/rte_distributor_priv.h:73:
warning: Found unknown command `\rte_distributor_process'

Regards,
/Bruce

Re: [dpdk-dev] [PATCH v4] mempool: use cache in single producer or consumer mode

[Olivier Matz]

On Wed, 11 Jan 2017 02:25:28 +, Wenfeng Liu
 wrote:
> Currently we will check mempool flags when we put/get objects from
> mempool. However, this makes cache useless when mempool is SC|SP,
> SC|MP, MC|SP cases.
> This patch makes cache available in above cases and improves
> performance.
> 
> Signed-off-by: Wenfeng Liu 

Acked-by: Olivier Matz 

[dpdk-dev] [PATCH v5 2/3] crypto: add sgl support in openssl PMD

[Tomasz Kulasek]

Signed-off-by: Tomasz Kulasek 
---
 doc/guides/cryptodevs/openssl.rst|3 +-
 drivers/crypto/openssl/rte_openssl_pmd.c |  329 +++---
 2 files changed, 259 insertions(+), 73 deletions(-)

diff --git a/doc/guides/cryptodevs/openssl.rst 
b/doc/guides/cryptodevs/openssl.rst
index f1c39ba..f6ed6ea 100644
--- a/doc/guides/cryptodevs/openssl.rst
+++ b/doc/guides/cryptodevs/openssl.rst
@@ -112,6 +112,7 @@ Limitations
 ---
 
 * Maximum number of sessions is 2048.
-* Chained mbufs are not supported.
+* Chained mbufs are supported only for source mbuf (destination must be
+  contiguous).
 * Hash only is not supported for GCM and GMAC.
 * Cipher only is not supported for GCM and GMAC.
diff --git a/drivers/crypto/openssl/rte_openssl_pmd.c 
b/drivers/crypto/openssl/rte_openssl_pmd.c
index 312154a..426e407 100644
--- a/drivers/crypto/openssl/rte_openssl_pmd.c
+++ b/drivers/crypto/openssl/rte_openssl_pmd.c
@@ -484,24 +484,112 @@
  * Process Operations
  
*--
  */
+static inline int
+process_openssl_encryption_update(struct rte_mbuf *mbuf_src, int offset,
+   uint8_t **dst, int srclen, EVP_CIPHER_CTX *ctx)
+{
+   struct rte_mbuf *m;
+   int dstlen;
+   int l, n = srclen;
+   uint8_t *src;
+
+   for (m = mbuf_src; m != NULL && offset > rte_pktmbuf_data_len(m);
+   m = m->next)
+   offset -= rte_pktmbuf_data_len(m);
+
+   if (m == 0)
+   return -1;
+
+   src = rte_pktmbuf_mtod_offset(m, uint8_t *, offset);
+
+   l = rte_pktmbuf_data_len(m) - offset;
+   if (srclen <= l) {
+   if (EVP_EncryptUpdate(ctx, *dst, &dstlen, src, srclen) <= 0)
+   return -1;
+   *dst += l;
+   return 0;
+   }
+
+   if (EVP_EncryptUpdate(ctx, *dst, &dstlen, src, l) <= 0)
+   return -1;
+
+   *dst += dstlen;
+   n -= l;
+
+   for (m = m->next; (m != NULL) && (n > 0); m = m->next) {
+   src = rte_pktmbuf_mtod(m, uint8_t *);
+   l = rte_pktmbuf_data_len(m) < n ? rte_pktmbuf_data_len(m) : n;
+   if (EVP_EncryptUpdate(ctx, *dst, &dstlen, src, l) <= 0)
+   return -1;
+   *dst += dstlen;
+   n -= l;
+   }
+
+   return 0;
+}
+
+static inline int
+process_openssl_decryption_update(struct rte_mbuf *mbuf_src, int offset,
+   uint8_t **dst, int srclen, EVP_CIPHER_CTX *ctx)
+{
+   struct rte_mbuf *m;
+   int dstlen;
+   int l, n = srclen;
+   uint8_t *src;
+
+   for (m = mbuf_src; m != NULL && offset > rte_pktmbuf_data_len(m);
+   m = m->next)
+   offset -= rte_pktmbuf_data_len(m);
+
+   if (m == 0)
+   return -1;
+
+   src = rte_pktmbuf_mtod_offset(m, uint8_t *, offset);
+
+   l = rte_pktmbuf_data_len(m) - offset;
+   if (srclen <= l) {
+   if (EVP_DecryptUpdate(ctx, *dst, &dstlen, src, srclen) <= 0)
+   return -1;
+   *dst += l;
+   return 0;
+   }
+
+   if (EVP_DecryptUpdate(ctx, *dst, &dstlen, src, l) <= 0)
+   return -1;
+
+   *dst += dstlen;
+   n -= l;
+
+   for (m = m->next; (m != NULL) && (n > 0); m = m->next) {
+   src = rte_pktmbuf_mtod(m, uint8_t *);
+   l = rte_pktmbuf_data_len(m) < n ? rte_pktmbuf_data_len(m) : n;
+   if (EVP_DecryptUpdate(ctx, *dst, &dstlen, src, l) <= 0)
+   return -1;
+   *dst += dstlen;
+   n -= l;
+   }
+
+   return 0;
+}
 
 /** Process standard openssl cipher encryption */
 static int
-process_openssl_cipher_encrypt(uint8_t *src, uint8_t *dst,
-   uint8_t *iv, uint8_t *key, int srclen,
+process_openssl_cipher_encrypt(struct rte_mbuf *mbuf_src, uint8_t *dst,
+   int offset, uint8_t *iv, uint8_t *key, int srclen,
EVP_CIPHER_CTX *ctx, const EVP_CIPHER *algo)
 {
-   int dstlen, totlen;
+   int totlen;
 
if (EVP_EncryptInit_ex(ctx, algo, NULL, key, iv) <= 0)
goto process_cipher_encrypt_err;
 
EVP_CIPHER_CTX_set_padding(ctx, 0);
 
-   if (EVP_EncryptUpdate(ctx, dst, &dstlen, src, srclen) <= 0)
+   if (process_openssl_encryption_update(mbuf_src, offset, &dst,
+   srclen, ctx))
goto process_cipher_encrypt_err;
 
-   if (EVP_EncryptFinal_ex(ctx, dst + dstlen, &totlen) <= 0)
+   if (EVP_EncryptFinal_ex(ctx, dst, &totlen) <= 0)
goto process_cipher_encrypt_err;
 
return 0;
@@ -513,23 +601,23 @@
 
 /** Process standard openssl cipher decryption */
 static int
-process_openssl_cipher_decrypt(uint8_t *src, uint8_t *dst,
-   uint8_t *iv, uint8_t *key, int srclen,
+process_openssl_cipher_decrypt(struct rte_mbuf *mbuf_src, uint8_t *dst,

[dpdk-dev] [PATCH v5 1/3] crypto: add sgl support in sw PMDs

[Tomasz Kulasek]

This patch introduces RTE_CRYPTODEV_FF_MBUF_SCATTER_GATHER feature flag
informing that selected crypto device supports segmented mbufs natively
and doesn't need to be coalesced before crypto operation.

While using segmented buffers in crypto devices may have unpredictable
results, for PMDs which doesn't support it natively, additional check is
made for debug compilation.

Signed-off-by: Tomasz Kulasek 
---
 drivers/crypto/aesni_gcm/aesni_gcm_pmd.c   |   14 ++
 drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c |   19 ---
 drivers/crypto/kasumi/rte_kasumi_pmd.c |   13 +
 drivers/crypto/null/null_crypto_pmd.c  |3 ++-
 drivers/crypto/snow3g/rte_snow3g_pmd.c |   15 +++
 drivers/crypto/zuc/rte_zuc_pmd.c   |   13 +
 lib/librte_cryptodev/rte_cryptodev.c   |4 ++--
 lib/librte_cryptodev/rte_cryptodev.h   |2 ++
 8 files changed, 77 insertions(+), 6 deletions(-)

diff --git a/drivers/crypto/aesni_gcm/aesni_gcm_pmd.c 
b/drivers/crypto/aesni_gcm/aesni_gcm_pmd.c
index af3d60f..5af22f7 100644
--- a/drivers/crypto/aesni_gcm/aesni_gcm_pmd.c
+++ b/drivers/crypto/aesni_gcm/aesni_gcm_pmd.c
@@ -377,6 +377,20 @@
break;
}
 
+#ifdef RTE_LIBRTE_PMD_AESNI_GCM_DEBUG
+   if (!rte_pktmbuf_is_contiguous(ops[i]->sym->m_src) ||
+   (ops[i]->sym->m_dst != NULL &&
+   !rte_pktmbuf_is_contiguous(
+   ops[i]->sym->m_dst))) {
+   ops[i]->status = RTE_CRYPTO_OP_STATUS_INVALID_ARGS;
+   GCM_LOG_ERR("PMD supports only contiguous mbufs, "
+   "op (%p) provides noncontiguous mbuf as "
+   "source/destination buffer.\n", ops[i]);
+   qp->qp_stats.enqueue_err_count++;
+   break;
+   }
+#endif
+
retval = process_gcm_crypto_op(qp, ops[i]->sym, sess);
if (retval < 0) {
ops[i]->status = RTE_CRYPTO_OP_STATUS_INVALID_ARGS;
diff --git a/drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c 
b/drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c
index 6d27d75..25f681b 100644
--- a/drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c
+++ b/drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c
@@ -571,15 +571,28 @@
int i, processed_jobs = 0;
 
for (i = 0; i < nb_ops; i++) {
-#ifdef RTE_LIBRTE_AESNI_MB_DEBUG
-   if (unlikely(op->type != RTE_CRYPTO_OP_TYPE_SYMMETRIC)) {
+#ifdef RTE_LIBRTE_PMD_AESNI_MB_DEBUG
+   if (unlikely(ops[i]->type != RTE_CRYPTO_OP_TYPE_SYMMETRIC)) {
MB_LOG_ERR("PMD only supports symmetric crypto "
"operation requests, op (%p) is not a "
-   "symmetric operation.", op);
+   "symmetric operation.", ops[i]);
+   qp->stats.enqueue_err_count++;
+   goto flush_jobs;
+   }
+
+   if (!rte_pktmbuf_is_contiguous(ops[i]->sym->m_src) ||
+   (ops[i]->sym->m_dst != NULL &&
+   !rte_pktmbuf_is_contiguous(
+   ops[i]->sym->m_dst))) {
+   MB_LOG_ERR("PMD supports only contiguous mbufs, "
+   "op (%p) provides noncontiguous mbuf as "
+   "source/destination buffer.\n", ops[i]);
+   ops[i]->status = RTE_CRYPTO_OP_STATUS_INVALID_ARGS;
qp->stats.enqueue_err_count++;
goto flush_jobs;
}
 #endif
+
sess = get_session(qp, ops[i]);
if (unlikely(sess == NULL)) {
qp->stats.enqueue_err_count++;
diff --git a/drivers/crypto/kasumi/rte_kasumi_pmd.c 
b/drivers/crypto/kasumi/rte_kasumi_pmd.c
index b119da2..4bdd7bb 100644
--- a/drivers/crypto/kasumi/rte_kasumi_pmd.c
+++ b/drivers/crypto/kasumi/rte_kasumi_pmd.c
@@ -455,6 +455,19 @@
for (i = 0; i < nb_ops; i++) {
curr_c_op = ops[i];
 
+#ifdef RTE_LIBRTE_PMD_KASUMI_DEBUG
+   if (!rte_pktmbuf_is_contiguous(curr_c_op->sym->m_src) ||
+   (curr_c_op->sym->m_dst != NULL &&
+   !rte_pktmbuf_is_contiguous(
+   curr_c_op->sym->m_dst))) {
+   KASUMI_LOG_ERR("PMD supports only contiguous mbufs, "
+   "op (%p) provides noncontiguous mbuf as "
+   "source/destination buffer.\n", curr_c_op);
+   curr_c_op->status = RTE_CRYPTO_OP_STATUS_INVALID_ARGS;
+   break;
+   }
+#endif
+
/* Set status as enqueued (not processed yet) by default. */