Re: [dpdk-dev] [PATCH 0/2] Multiple Pktmbuf mempool support

[Hemant Agrawal]

Hi Olivier,
Any opinion on this patchset?

Regards,
Hemant

On 7/4/2017 5:52 PM, Hemant Agrawal wrote:

This patch is in addition to the patch series[1] submitted by
Santosh to allow application to set mempool handle.

The existing pktmbuf pool create api only support the internal use
of "CONFIG_RTE_MBUF_DEFAULT_MEMPOOL_OPS", which assumes that the HW
can only support one type of mempool for packet mbuf.

There are multiple additional requirements.

1. The platform independent image detects the underlying bus,
based on the bus and resource detected, it will dynamically select
the default mempool. This need not to have the application knowlege.
e.g. DPAA2 and DPAA are two different NXP platforms, based on the
underlying platform the default ops for mbuf can be dpaa or dpaa2.
Application should work seemlessly whether it is running on dpaa or dpaa2.

2.Platform support more than one type of mempool for pktmbuf,
depend on the availability of resource, the driver can decide one
of the mempool for the current packet mbuf request.

3. In case of where application is providing the mempool, as proposed
in [1], the check preference logic will be bypassed and application
config will take priority.

[1]Allow application set mempool handle
http://dpdk.org/ml/archives/dev/2017-June/067022.html

Hemant Agrawal (2):
  mempool: check the support for the given mempool
  mbuf: add support for preferred mempool list

 config/common_base   |  2 ++
 lib/librte_mbuf/rte_mbuf.c   | 28 +++-
 lib/librte_mempool/rte_mempool.h | 24 
 lib/librte_mempool/rte_mempool_ops.c | 32 
 4 files changed, 81 insertions(+), 5 deletions(-)

Re: [dpdk-dev] [PATCH v3 21/40] maintainers: claim ownership of DPAA Mempool driver

[Shreyansh Jain]

On Friday 22 September 2017 12:23 PM, Thomas Monjalon wrote:

22/09/2017 08:47, Shreyansh Jain:

On Friday 22 September 2017 03:26 AM, Thomas Monjalon wrote:

23/08/2017 16:11, Shreyansh Jain:

--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -409,6 +409,7 @@ NXP dpaa
   M: Hemant Agrawal 
   M: Shreyansh Jain 
   F: drivers/bus/dpaa/
+F: drivers/mempool/dpaa/
   F: doc/guides/nics/dpaa.rst
   F: doc/guides/nics/features/dpaa.ini


This kind of patch can be squashed in the first patch introducing
this new directory.



Then the patch script (devtools/check-git-log.sh) reports error - I
think. That is the primary reason I split them across multiple patches.
You sure that doesn't matter?


Which error?

To be clear I suggest to squash with patch 19 where
drivers/mempool/dpaa/Makefile is introduced.



Yes, I understand that.
It would report error that the headline is wrong because I am hitting 
different directories - "MAINTAINERS" and "drivers/mempool/*" with the 
same patch having headline "mempool/*".

Re: [dpdk-dev] [PATCH v3 4/6] net/i40e: add new functions to manipulate with pctype mapping table

[Xing, Beilei]

> -Original Message-
> From: Rybalchenko, Kirill
> Sent: Wednesday, September 20, 2017 10:33 PM
> To: dev@dpdk.org
> Cc: Rybalchenko, Kirill ; Chilikin, Andrey
> ; Xing, Beilei ; Wu,
> Jingjing 
> Subject: [PATCH v3 4/6] net/i40e: add new functions to manipulate with
> pctype mapping table
> 
> Add new functions which allow modify, return or reset to default the
> contents of flow type to pctype dynamic mapping table.
> 
> v3:
> function rte_pmd_i40e_flow_type_mapping_get returns now full mapping
> table.
> 
> Signed-off-by: Kirill Rybalchenko 
> ---


> diff --git a/drivers/net/i40e/rte_pmd_i40e_version.map
> b/drivers/net/i40e/rte_pmd_i40e_version.map
> index ef8882b..438ca81 100644
> --- a/drivers/net/i40e/rte_pmd_i40e_version.map
> +++ b/drivers/net/i40e/rte_pmd_i40e_version.map
> @@ -50,5 +50,8 @@ DPDK_17.11 {
>   global:
> 
>   rte_pmd_i40e_add_vf_mac_addr;
> + rte_pmd_i40e_flow_type_mapping_update;
> + rte_pmd_i40e_flow_type_mapping_get;
> + rte_pmd_i40e_flow_type_mapping_get;

Should it be rte_pmd_i40e_flow_type_mapping_reset here?

> 
>  } DPDK_17.08;
> --
> 2.5.5

Re: [dpdk-dev] [PATCH v3 2/6] net/i40e: add definition for invalid pctype

[Xing, Beilei]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Kirill Rybalchenko
> Sent: Wednesday, September 20, 2017 10:33 PM
> To: dev@dpdk.org
> Cc: Rybalchenko, Kirill ; Chilikin, Andrey
> ; Xing, Beilei ; Wu,
> Jingjing 
> Subject: [dpdk-dev] [PATCH v3 2/6] net/i40e: add definition for invalid
> pctype
> 
> Add new definition in enum i40e_filter_pctype for for invalid pctype
> 
> Signed-off-by: Kirill Rybalchenko 
> ---
>  drivers/net/i40e/base/i40e_type.h | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/i40e/base/i40e_type.h
> b/drivers/net/i40e/base/i40e_type.h
> index dca725a..d57986e 100644
> --- a/drivers/net/i40e/base/i40e_type.h
> +++ b/drivers/net/i40e/base/i40e_type.h
> @@ -1245,9 +1245,11 @@ struct i40e_filter_program_desc {
> 
>  /* Packet Classifier Types for filters */  enum i40e_filter_pctype {
> - /* Note: Values 0-28 are reserved for future use.
> + /* Note: Vlue 0 is not valid pctype.
> +  * Values 0-28 are reserved for future use.
>* Value 29, 30, 32 are not supported on XL710 and X710.
>*/
> + I40E_FILTER_PCTYPE_INVALID  = 0,

We'd better not to change base code which is aligned with kernel driver.
We can add macro in i40e_ethdev.h.

>   I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP= 29,
>   I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV4_UDP  = 30,
>   I40E_FILTER_PCTYPE_NONF_IPV4_UDP= 31,
> --
> 2.5.5

[dpdk-dev] [PATCH] net/i40e: fix uninitialized variable

[Yong Wang]

In func i40evf_dev_link_update(), "new_link.link_autoneg" is used in
func call i40evf_dev_atomic_write_link_status(), but is uninitialized.

Signed-off-by: Yong Wang 
---
 drivers/net/i40e/i40e_ethdev_vf.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/i40e/i40e_ethdev_vf.c 
b/drivers/net/i40e/i40e_ethdev_vf.c
index 73c315a..3c0f436 100644
--- a/drivers/net/i40e/i40e_ethdev_vf.c
+++ b/drivers/net/i40e/i40e_ethdev_vf.c
@@ -2218,6 +2218,7 @@ static int eth_i40evf_pci_remove(struct rte_pci_device 
*pci_dev)
new_link.link_duplex = ETH_LINK_FULL_DUPLEX;
new_link.link_status = vf->link_up ? ETH_LINK_UP :
 ETH_LINK_DOWN;
+   new_link.link_autoneg = ETH_LINK_AUTONEG;
 
i40evf_dev_atomic_write_link_status(dev, &new_link);
 
-- 
1.8.3.1

Re: [dpdk-dev] [PATCH v3 21/40] maintainers: claim ownership of DPAA Mempool driver

[Thomas Monjalon]

22/09/2017 09:37, Shreyansh Jain:
> On Friday 22 September 2017 12:23 PM, Thomas Monjalon wrote:
> > 22/09/2017 08:47, Shreyansh Jain:
> >> On Friday 22 September 2017 03:26 AM, Thomas Monjalon wrote:
> >>> 23/08/2017 16:11, Shreyansh Jain:
>  --- a/MAINTAINERS
>  +++ b/MAINTAINERS
>  @@ -409,6 +409,7 @@ NXP dpaa
> M: Hemant Agrawal 
> M: Shreyansh Jain 
> F: drivers/bus/dpaa/
>  +F: drivers/mempool/dpaa/
> F: doc/guides/nics/dpaa.rst
> F: doc/guides/nics/features/dpaa.ini
> >>>
> >>> This kind of patch can be squashed in the first patch introducing
> >>> this new directory.
> >>
> >> Then the patch script (devtools/check-git-log.sh) reports error - I
> >> think. That is the primary reason I split them across multiple patches.
> >> You sure that doesn't matter?
> > 
> > Which error?
> > 
> > To be clear I suggest to squash with patch 19 where
> > drivers/mempool/dpaa/Makefile is introduced.
> 
> Yes, I understand that.
> It would report error that the headline is wrong because I am hitting 
> different directories - "MAINTAINERS" and "drivers/mempool/*" with the 
> same patch having headline "mempool/*".

The test you are talking about has this comment:
# check headline prefix when touching only drivers, e.g. net/
If you hit a warning, there is a bug.

Re: [dpdk-dev] [PATCH v2 1/2] net/i40e: fix clear xstats bug in vf port

[Zhao1, Wei]

Hi,  Ferruh

> -Original Message-
> From: Yigit, Ferruh
> Sent: Friday, September 22, 2017 5:01 AM
> To: Zhao1, Wei ; dev@dpdk.org
> Cc: Wu, Jingjing 
> Subject: Re: [dpdk-dev] [PATCH v2 1/2] net/i40e: fix clear xstats bug in vf
> port
> 
> On 9/21/2017 7:16 PM, Ferruh Yigit wrote:
> > On 9/21/2017 4:11 AM, Zhao1, Wei wrote:
> >> Hi，Ferruh
> >>
> >>> -Original Message-
> >>> From: Yigit, Ferruh
> >>> Sent: Thursday, September 14, 2017 9:31 PM
> >>> To: Zhao1, Wei ; dev@dpdk.org
> >>> Cc: Wu, Jingjing 
> >>> Subject: Re: [dpdk-dev] [PATCH v2 1/2] net/i40e: fix clear xstats
> >>> bug in vf port
> >>>
> >>> On 9/1/2017 3:30 AM, Zhao1, Wei wrote:
>  Hi,  Ferruh
> 
> > -Original Message-
> > From: Yigit, Ferruh
> > Sent: Friday, September 1, 2017 12:54 AM
> > To: Zhao1, Wei ; dev@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH v2 1/2] net/i40e: fix clear xstats
> > bug in vf port
> >
> > On 8/29/2017 3:28 AM, Wei Zhao wrote:
> >> There is a bug in vf clear xstats command, it do not record the
> >> statics data in offset struct member.So, vf need to keep record
> >> of xstats data from pf and update the statics according to offset.
> >>
> >> Fixes: da61cd0849766 ("i40evf: add extended stats")
> >>
> >> Signed-off-by: Wei Zhao 
> >>
> >> ---
> >>
> >> Changes in v2:
> >>
> >>  fix patch log check warning.
> >> ---
> >>  app/test-pmd/config.c |  6 ++--
> >>  drivers/net/i40e/i40e_ethdev_vf.c | 64
> >> ++-
> >>  2 files changed, 67 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c index
> >> 3ae3e1c..14131d6 100644
> >> --- a/app/test-pmd/config.c
> >> +++ b/app/test-pmd/config.c
> >> @@ -203,8 +203,10 @@ nic_stats_display(portid_t port_id)
> >>if (diff_cycles > 0)
> >>diff_cycles = prev_cycles[port_id] - diff_cycles;
> >>
> >> -  diff_pkts_rx = stats.ipackets - prev_pkts_rx[port_id];
> >> -  diff_pkts_tx = stats.opackets - prev_pkts_tx[port_id];
> >> +  diff_pkts_rx = (stats.ipackets > prev_pkts_rx[port_id]) ?
> >> +  (stats.ipackets - prev_pkts_rx[port_id]) : 0;
> >> +  diff_pkts_tx = (stats.opackets > prev_pkts_tx[port_id]) ?
> >> +  (stats.opackets - prev_pkts_tx[port_id]) : 0;
> >
> > I guess this testpmd update is not directly related to this patch,
> > but to protect testpmd against value overflow? Can this be another
> patch?
> 
>  Nonono, this code change is directly related to this patch, if we
>  do not do this code change, the diff_pkts_rx and diff_pkts_tx
>  statistic data will
> >>> be wrong  when the first time after clear xstats command.
> >>>
> >>> If this testpmd code is only wrong for i40e vf after this patch,
> >>> perhaps something else is wrong? Perhaps we should update i40e vf
> stats.
> >>>
> >>> OR, if this code is already wrong, lets move it to its own patch.
> >>>
> >>
> >> A new patch will be commit later.
> >>
> 
> >
> > <...>
> >
> >>  static int
> >>  i40evf_get_statistics(struct rte_eth_dev *dev, struct
> >> rte_eth_stats
> >> *stats)  {
> >>int ret;
> >>struct i40e_eth_stats *pstats = NULL;
> >> +  struct i40e_vf *vf = I40EVF_DEV_PRIVATE_TO_VF(dev-
> >data-
> >> dev_private);
> >> +  struct i40e_vsi *vsi = &vf->vsi;
> >>
> >>ret = i40evf_update_stats(dev, &pstats);
> >>if (ret != 0)
> >>return 0;
> >>
> >> +  i40evf_update_vsi_stats(vsi, pstats);
> >
> > But not having this previously means all VF stats were wrong
> > previously, not only extended ones, also basic ones. And not not
> > wrong with small difference, this should give a big difference in the
> stats.
> >
> > I am suspicious about this part, because if this is the case, I
> > would expect this should be detected earlier.
> >
> > I have not traced the code, but is there any chance that
> >>> "eth_stats_offset"
> > has been used by other end of the admin command?
> 
>  To be frankly speaking, this bug is firstly discovered by a big user.
>  This bug only appear after use CLI "clear port xstats 0". So it is
>  not easy to
> >>> detect this bug.
>  After using this fix patch ,the big user who report this issue has
>  feed back it
> >>> work well now.
>  The root cause is not so complicated, when the pf which admin this
>  vf is in kernel state, DPDK can not Give pf the info to clear and
>  update offset command, so vf can only keep record the offset data
>  in DPDK VF
> >>> port locally.
> >>>
> >>> Please help me understand this.
> >>>
> >>> 1- The problem you are fixing only seen with Linux PF, with DPDK PF
> >>> you don't see the proble

[dpdk-dev] [PATCH v2 2/5] eal/ppc64: define architecture specific rdtsc hz

[Gowrishankar]

From: Jerin Jacob 

In ppc_64, rte_rdtsc() returns timebase register value which increments
at independent timebase frequency and hence not related to lcore cpu
frequency to derive TSC hz. Hence, we stick with master lcore frequency.

CC: Chao Zhu 
Signed-off-by: Jerin Jacob 
Signed-off-by: Gowrishankar Muthukrishnan 
---

v2:
 * add ppc_64 specific implementation

 .../common/include/arch/ppc_64/rte_cycles.h| 24 ++
 1 file changed, 24 insertions(+)

diff --git a/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h 
b/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h
index 8fa6fc6..1b36587 100644
--- a/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h
+++ b/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h
@@ -38,9 +38,13 @@
 #endif
 
 #include "generic/rte_cycles.h"
+#include "../../lib/librte_eal/common/eal_filesystem.h"
 
 #include 
 #include 
+#include 
+
+static const char sys_cpu_dir[] = "/sys/devices/system/cpu";
 
 /**
  * Read the time base register.
@@ -79,6 +83,26 @@
return tsc.tsc_64;
 }
 
+/**
+ * Get the number of rdtsc cycles in one second.
+ *
+ * @return
+ *   The number of rdtsc cycles in one second.
+ */
+static inline uint64_t
+rte_rdtsc_arch_hz(void)
+{
+   unsigned long cpu_hz;
+   char path[PATH_MAX];
+
+   snprintf(path, sizeof(path), "%s/cpu%d/cpufreq/cpuinfo_cur_freq",
+   sys_cpu_dir, rte_get_master_lcore());
+   if (eal_parse_sysfs_value(path, &cpu_hz) < 0)
+   RTE_LOG(WARNING, EAL, "Unable to parse %s\n", path);
+
+   return cpu_hz*1000;
+}
+
 static inline uint64_t
 rte_rdtsc_precise(void)
 {
-- 
1.9.1

[dpdk-dev] [PATCH v2 0/5] improve tsc frequency calibration

[Gowrishankar]

From: Gowrishankar Muthukrishnan 

Some architecture like armv8 provides an architecture specific function
to get the rdtsc frequency. The existing rdtsc calibration scheme uses
OS serivce like sleep(1) to calibrate the frequency which may not
produce the accurate result. Introducing an architecture specific hook
to get the rdtsc frequency if architecture provides it. If not, use the
exiting the calibrate scheme to get the rdtsc frequency.

Jerin Jacob (5):
  eal/x86: define architecture specific rdtsc hz
  eal/ppc64: define architecture specific rdtsc hz
  eal/armv7: define architecture specific rdtsc hz
  eal/armv8: define architecture specific rdtsc hz
  eal/timer: honor architecture specific rdtsc hz function

 lib/librte_eal/common/eal_common_timer.c   |  5 +++-
 .../common/include/arch/arm/rte_cycles_32.h| 13 ++
 .../common/include/arch/arm/rte_cycles_64.h| 30 ++
 .../common/include/arch/ppc_64/rte_cycles.h| 24 +
 .../common/include/arch/x86/rte_cycles.h   | 13 ++
 5 files changed, 84 insertions(+), 1 deletion(-)

-- 
1.9.1

[dpdk-dev] [PATCH v2 4/5] eal/armv8: define architecture specific rdtsc hz

[Gowrishankar]

From: Jerin Jacob 

Use cntvct_el0 system register to get the system counter frequency.

If the system is configured with RTE_ARM_EAL_RDTSC_USE_PMU then
return 0(let the common code calibrate the tsc frequency).

CC: Jianbo Liu 
Signed-off-by: Jerin Jacob 
Acked-by: Jianbo Liu 
---
 .../common/include/arch/arm/rte_cycles_64.h| 30 ++
 1 file changed, 30 insertions(+)

diff --git a/lib/librte_eal/common/include/arch/arm/rte_cycles_64.h 
b/lib/librte_eal/common/include/arch/arm/rte_cycles_64.h
index 1545769..5b95cb6 100644
--- a/lib/librte_eal/common/include/arch/arm/rte_cycles_64.h
+++ b/lib/librte_eal/common/include/arch/arm/rte_cycles_64.h
@@ -58,6 +58,23 @@
asm volatile("mrs %0, cntvct_el0" : "=r" (tsc));
return tsc;
 }
+
+/**
+ * Get the number of rdtsc cycles in one second if the architecture supports.
+ *
+ * @return
+ *   The number of rdtsc cycles in one second. Return zero if the architecture
+ *   support is not available.
+ */
+static inline uint64_t
+rte_rdtsc_arch_hz(void)
+{
+   uint64_t freq;
+
+   asm volatile("mrs %0, cntfrq_el0" : "=r" (freq));
+   return freq;
+}
+
 #else
 /**
  * This is an alternative method to enable rte_rdtsc() with high resolution
@@ -85,6 +102,19 @@
asm volatile("mrs %0, pmccntr_el0" : "=r"(tsc));
return tsc;
 }
+
+/**
+ * Get the number of rdtsc cycles in one second if the architecture supports.
+ *
+ * @return
+ *   The number of rdtsc cycles in one second. Return zero if the architecture
+ *   support is not available.
+ */
+static inline uint64_t
+rte_rdtsc_arch_hz(void)
+{
+   return 0;
+}
 #endif
 
 static inline uint64_t
-- 
1.9.1

[dpdk-dev] [PATCH v2 5/5] eal/timer: honor architecture specific rdtsc hz function

[Gowrishankar]

From: Jerin Jacob 

When calibrating the tsc frequency, first, probe the architecture specific
rdtsc hz function. if not available, use the existing calibrate scheme
to calibrate the tsc frequency.

Signed-off-by: Jerin Jacob 
---
 lib/librte_eal/common/eal_common_timer.c | 5 -
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/lib/librte_eal/common/eal_common_timer.c 
b/lib/librte_eal/common/eal_common_timer.c
index ed0b16d..978edae 100644
--- a/lib/librte_eal/common/eal_common_timer.c
+++ b/lib/librte_eal/common/eal_common_timer.c
@@ -80,8 +80,11 @@
 void
 set_tsc_freq(void)
 {
-   uint64_t freq = get_tsc_freq();
+   uint64_t freq;
 
+   freq = rte_rdtsc_arch_hz();
+   if (!freq)
+   freq = get_tsc_freq();
if (!freq)
freq = estimate_tsc_freq();
 
-- 
1.9.1

[dpdk-dev] [PATCH v2 3/5] eal/armv7: define architecture specific rdtsc hz

[Gowrishankar]

From: Jerin Jacob 

CC: Jan Viktorin 
CC: Jianbo Liu 
Signed-off-by: Jerin Jacob 
---
 lib/librte_eal/common/include/arch/arm/rte_cycles_32.h | 13 +
 1 file changed, 13 insertions(+)

diff --git a/lib/librte_eal/common/include/arch/arm/rte_cycles_32.h 
b/lib/librte_eal/common/include/arch/arm/rte_cycles_32.h
index 9c1be71..68d7462 100644
--- a/lib/librte_eal/common/include/arch/arm/rte_cycles_32.h
+++ b/lib/librte_eal/common/include/arch/arm/rte_cycles_32.h
@@ -104,6 +104,19 @@
 
 #endif /* RTE_ARM_EAL_RDTSC_USE_PMU */
 
+/**
+ * Get the number of rdtsc cycles in one second if the architecture supports.
+ *
+ * @return
+ *   The number of rdtsc cycles in one second. Return zero if the architecture
+ *   support is not available.
+ */
+static inline uint64_t
+rte_rdtsc_arch_hz(void)
+{
+   return 0;
+}
+
 static inline uint64_t
 rte_rdtsc_precise(void)
 {
-- 
1.9.1

[dpdk-dev] [PATCH v2 1/5] eal/x86: define architecture specific rdtsc hz

[Gowrishankar]

From: Jerin Jacob 

CC: Bruce Richardson 
CC: Konstantin Ananyev 

Signed-off-by: Jerin Jacob 
Acked-by: Anatoly Burakov 
---
 lib/librte_eal/common/include/arch/x86/rte_cycles.h | 13 +
 1 file changed, 13 insertions(+)

diff --git a/lib/librte_eal/common/include/arch/x86/rte_cycles.h 
b/lib/librte_eal/common/include/arch/x86/rte_cycles.h
index 1bb3e1d..e2661e2 100644
--- a/lib/librte_eal/common/include/arch/x86/rte_cycles.h
+++ b/lib/librte_eal/common/include/arch/x86/rte_cycles.h
@@ -77,6 +77,19 @@
return tsc.tsc_64;
 }
 
+/**
+ * Get the number of rdtsc cycles in one second if the architecture supports.
+ *
+ * @return
+ *   The number of rdtsc cycles in one second. Return zero if the architecture
+ *   support is not available.
+ */
+static inline uint64_t
+rte_rdtsc_arch_hz(void)
+{
+   return 0;
+}
+
 static inline uint64_t
 rte_rdtsc_precise(void)
 {
-- 
1.9.1

Re: [dpdk-dev] [PATCH v2] librte_eal: fix wrong assert for arm and ppc

[Jan Viktorin]

On Thu, 21 Sep 2017 17:59:53 +0200
Lukasz Majczak  wrote:

> The assertion of return value from the open() function is done against
> 0, while it is a correct value - open() returns -1 in case of an error.
> It causes problems while trying to run as a daemon, in which case, this
> call to open() will return 0 as a valid descriptor.

I'd put an empty line here...

> Fixes: b94e5c9406b5 ("eal/arm: add CPU flags for ARMv7")
> Fixes: 97523f822ba9 ("eal/arm: add CPU flags for ARMv8")
> Fixes: 9ae155385686 ("eal/ppc: cpu flag checks for IBM Power")
> 
> Signed-off-by: Lukasz Majczak 

Acked-by: Jan Viktorin 

Re: [dpdk-dev] [PATCH] hash: fix incorrect eviction counter

[Bruce Richardson]

On Thu, Sep 21, 2017 at 01:46:46PM +0100, Pablo de Lara wrote:
> When adding a new entry in a hash table, there is
> a maximum number of evictions that can be
> performed. When the counter of these evictions reaches
> this maximum, the entry cannot be added, as it is considered
> that the algorithm has encountered an infinite loop.
> 
> The problem with the current implementation, is that this
> counter was declared as a static variable.
> If there are multiple threads adding entries in the same table
> or in different tables, they should access different counters,
> one per core and per table.
> 
> Therefore, an array of counter has been added to the
> hash table structure.
> 
> Fixes: 243e93a5046f ("hash: fix unlimited cuckoo path")
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Pablo de Lara 
> ---
Since you appear to be passing this counter through the different
function calls as parameter, and it gets reset to zero at the start, do
you even need an array. Can you not just have a non-static local
variable in the function?

/Bruce

Re: [dpdk-dev] [PATCH v3 1/2] net/i40e: queue region set and flush

[Zhao1, Wei]

Hi，Andrey

> -Original Message-
> From: Chilikin, Andrey
> Sent: Friday, September 22, 2017 3:53 AM
> To: Zhao1, Wei ; dev@dpdk.org
> Cc: Zhao1, Wei ; Yigit, Ferruh
> ; Xing, Beilei ; Wu, Jingjing
> 
> Subject: RE: [dpdk-dev] [PATCH v3 1/2] net/i40e: queue region set and flush
> 
> > -Original Message-
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Wei Zhao
> > Sent: Friday, September 15, 2017 4:13 AM
> > To: dev@dpdk.org
> > Cc: Zhao1, Wei 
> > Subject: [dpdk-dev] [PATCH v3 1/2] net/i40e: queue region set and
> > flush
> >
> > This feature enable queue regions configuration for RSS in PF/VF, so
> > that different traffic classes or different packet classification
> > types can be separated to different queues in different queue
> > regions.This patch can set queue region range, it include queue number
> > in a region and the index of first queue.
> > This patch enable mapping between different priorities (UP) and
> > different traffic classes.It also enable mapping between a region
> > index and a sepcific flowtype(PCTYPE).It also provide the solution of
> > flush all configuration about queue region the above described.
> >
> > Signed-off-by: Wei Zhao 
> > ---
> >  drivers/net/i40e/i40e_ethdev.c|  19 +-
> >  drivers/net/i40e/i40e_ethdev.h|  30 ++
> >  drivers/net/i40e/rte_pmd_i40e.c   | 482
> > ++
> >  drivers/net/i40e/rte_pmd_i40e.h   |  38 +++
> >  drivers/net/i40e/rte_pmd_i40e_version.map |   1 +
> >  5 files changed, 566 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/net/i40e/i40e_ethdev.c
> > b/drivers/net/i40e/i40e_ethdev.c index 8e0580c..1663fc0 100644
> > --- a/drivers/net/i40e/i40e_ethdev.c
> > +++ b/drivers/net/i40e/i40e_ethdev.c
> > @@ -137,10 +137,6 @@
> >  #define I40E_PRTTSYN_TSYNTYPE0x0e00
> >  #define I40E_CYCLECOUNTER_MASK   0xULL
> >
> > -#define I40E_MAX_PERCENT100
> > -#define I40E_DEFAULT_DCB_APP_NUM1
> > -#define I40E_DEFAULT_DCB_APP_PRIO   3
> > -
> >  /**
> >   * Below are values for writing un-exposed registers suggested
> >   * by silicon experts
> > @@ -1034,6 +1030,15 @@ i40e_init_fdir_filter_list(struct rte_eth_dev
> *dev)
> > return ret;
> >  }
> >
> > +static void
> > +i40e_init_queue_region_conf(struct rte_eth_dev *dev) {
> > +   struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data-
> > >dev_private);
> > +   struct i40e_queue_region_info *info = &pf->queue_region;
> > +
> > +   memset(info, 0, sizeof(struct i40e_queue_region_info)); }
> > +
> >  static int
> >  eth_i40e_dev_init(struct rte_eth_dev *dev)  { @@ -1309,6 +1314,9 @@
> > eth_i40e_dev_init(struct rte_eth_dev *dev)
> > if (ret < 0)
> > goto err_init_fdir_filter_list;
> >
> > +   /* initialize queue region configuration */
> > +   i40e_init_queue_region_conf(dev);
> > +
> > return 0;
> >
> >  err_init_fdir_filter_list:
> > @@ -1464,6 +1472,9 @@ eth_i40e_dev_uninit(struct rte_eth_dev *dev)
> > /* Remove all Traffic Manager configuration */
> > i40e_tm_conf_uninit(dev);
> >
> > +   /* Remove all the queue region configuration */
> > +   i40e_flush_region_all_conf(hw, pf, 0);
> > +
> > return 0;
> >  }
> >
> > diff --git a/drivers/net/i40e/i40e_ethdev.h
> > b/drivers/net/i40e/i40e_ethdev.h index ad80f0f..d612886 100644
> > --- a/drivers/net/i40e/i40e_ethdev.h
> > +++ b/drivers/net/i40e/i40e_ethdev.h
> > @@ -260,6 +260,16 @@ enum i40e_flxpld_layer_idx {  #define
> > I40E_QOS_BW_WEIGHT_MIN 1
> >  /* The max bandwidth weight is 127. */  #define
> > I40E_QOS_BW_WEIGHT_MAX 127
> > +/* The max queue region index is 7. */ #define
> I40E_REGION_MAX_INDEX
> > +7
> > +/* The max queue region userpriority is 7. */ #define
> > +I40E_REGION_USERPRIORITY_MAX_INDEX 7
> > +/* The max pctype index is 63. */
> > +#define I40E_REGION_PCTYPE_MAX_INDEX 63
> > +
> > +#define I40E_MAX_PERCENT100
> > +#define I40E_DEFAULT_DCB_APP_NUM1
> > +#define I40E_DEFAULT_DCB_APP_PRIO   3
> >
> >  /**
> >   * The overhead from MTU to max frame size.
> > @@ -541,6 +551,22 @@ struct i40e_ethertype_rule {
> > struct rte_hash *hash_table;
> >  };
> >
> > +/* queue region info */
> > +struct i40e_region_info {
> > +   uint8_t region_id;
> > +   uint8_t queue_start_index;
> > +   uint8_t queue_num;
> > +   uint8_t user_priority_num;
> > +   uint8_t user_priority[I40E_MAX_USER_PRIORITY];
> > +   uint8_t flowtype_num;
> > +   uint8_t hw_flowtype[I40E_FILTER_PCTYPE_MAX];
> > +};
> > +
> > +struct i40e_queue_region_info {
> > +   uint16_t queue_region_number;
> > +   struct i40e_region_info region[I40E_REGION_MAX_INDEX + 1]; };
> > +
> >  /* Tunnel filter number HW supports */  #define
> > I40E_MAX_TUNNEL_FILTER_NUM 400
> >
> > @@ -776,6 +802,7 @@ struct i40e_pf {
> > struct i40e_fdir_info fdir; /* flow director info */
> > struct i40e_ethertype_rule ethertype; /* Ethertype filter rule */
> > struct i40e_tunnel_rule tunnel; /* Tunnel filter rule */
> 

Re: [dpdk-dev] [PATCH v4 3/4] eventdev: Add eventdev ethernet Rx adapter

[Jerin Jacob]

-Original Message-
> Date: Fri, 22 Sep 2017 02:47:13 +0530
> From: Nikhil Rao 
> To: jerin.ja...@caviumnetworks.com, bruce.richard...@intel.com
> CC: gage.e...@intel.com, dev@dpdk.org, tho...@monjalon.net,
>  harry.van.haa...@intel.com, hemant.agra...@nxp.com, nipun.gu...@nxp.com,
>  narender.vang...@intel.com, erik.g.carri...@intel.com,
>  abhinandan.guj...@intel.com, santosh.shu...@caviumnetworks.com
> Subject: [PATCH v4 3/4] eventdev: Add eventdev ethernet Rx adapter
> X-Mailer: git-send-email 2.7.4
> 
> Add common APIs for configuring packet transfer from ethernet Rx
> queues to event devices across HW & SW packet transfer mechanisms.
> A detailed description of the adapter is contained in the header's
> comments.
> 
> The adapter implementation uses eventdev PMDs to configure the packet
> transfer if HW support is available and if not, it uses an EAL service
> function that reads packets from ethernet Rx queues and injects these
> as events into the event device.
> 
> Signed-off-by: Nikhil Rao 
> Signed-off-by: Gage Eads 
> Signed-off-by: Abhinandan Gujjar 

Overall it looks good. A few top-level comments to start with,

1) Please split this patch to minimum two
a) Specification header
b) and the implementation

2) Please add a new section in MAINTAINERS files and update the new files
responsibility.

3) The doxygen file is not hooked into the documentation build.
Check the doc/api/doxy-api-index.md file. You can use "make doc-api-html"
to verify the doxygen html output.

4) Since the APIs looks good and if there is no other objection,
Can you add a programmer guide for Rx adapter.
If you are busy it is fine not have in next version. Post RC1 or RC2 is
fine. What do you think?


> ---
>  lib/librte_eventdev/rte_event_eth_rx_adapter.h |  384 
>  lib/librte_eventdev/rte_event_eth_rx_adapter.c | 1238 
> 
>  lib/Makefile   |2 +-
>  lib/librte_eventdev/Makefile   |2 +
>  lib/librte_eventdev/rte_eventdev_version.map   |   11 +-
>  5 files changed, 1635 insertions(+), 2 deletions(-)
>  create mode 100644 lib/librte_eventdev/rte_event_eth_rx_adapter.h
>  create mode 100644 lib/librte_eventdev/rte_event_eth_rx_adapter.c
> 
> +#ifndef _RTE_EVENT_ETH_RX_ADAPTER_
> +#define _RTE_EVENT_ETH_RX_ADAPTER_
> +
> +/**
> + * @file
> + *
> + * RTE Event Ethernet Rx Adapter
> + *
> + * An eventdev-based packet processing application enqueues/dequeues mbufs
> + * to/from the event device. The application uses the adapter APIs to 
> configure
> + * the packet flow between the ethernet devices and event devices. Depending 
> on
> + * on the capabilties of the eventdev PMD, the adapter may use a EAL service

s/capabilties/capabilities

> + * core function for packet transfer or use internal PMD functions to 
> configure
> + * the packet transfer between the ethernet device and the event device.
> + *
> + * The ethernet Rx event adapter's functions are:
> + *  - rte_event_eth_rx_adapter_create_ext()
> + *  - rte_event_eth_rx_adapter_create()/free()
> + *  - rte_event_eth_rx_adapter_queue_add()/del()
> + *  - rte_event_eth_rx_adapter_start()/stop()
> + *  - rte_event_eth_rx_adapter_stats_get()/reset()
> + *
> + * The applicaton creates an event to ethernet adapter using

How about,
The application creates an ethernet device to event device adapter using

> + * rte_event_eth_rx_adapter_create_ext() or rte_event_eth_rx_adapter_create()
> + * functions.
> + * The adapter needs to know which ethernet rx queues to poll for mbufs as 
> well
> + * as event device parameters such as the event queue identifier, event
> + * priority and scheduling type that the adapter should use when constructing
> + * events. The rte_event_eth_rx_adapter_queue_add() function is provided for
> + * this purpose.
> + * The servicing weight parameter in the rte_event_eth_rx_adapter_queue_conf
> + * is applicable when the Rx adapter uses a service core function and is
> + * intended to provide application control of the polling frequency of 
> ethernet
> + * device receive queues, for example, the application may want to poll 
> higher
> + * priority queues with a higher frequency but at the same time not starve
> + * lower priority queues completely. If this parameter is zero and the 
> receive
> + * interrupt is enabled when configuring the device, the receive queue is
> + * interrupt driven; else, the queue is assigned a servicing weight of one.
> + *
> + * If the adapter uses a rte_service function, then the application is also
> + * required to assign a core to the service function and control the service
> + * core using the rte_service APIs. The 
> rte_event_eth_rx_adapter_service_id_get
> + * function can be used to retrieve the service function ID of the adapter in
> + * this case.
> + *
> + * Note: Interrupt driven receive queues are currentely unimplemented.

s/currentely/currently

> + */
> +
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +#incl

[dpdk-dev] [PATCH v3 1/8] mbuf: support GTP in software packet type parser

[Beilei Xing]

Add support of GTP-C and GTP-U tunnels in rte_net_get_ptype().

Signed-off-by: Beilei Xing 
---
 lib/librte_mbuf/rte_mbuf_ptype.c |  2 ++
 lib/librte_mbuf/rte_mbuf_ptype.h | 24 
 2 files changed, 26 insertions(+)

diff --git a/lib/librte_mbuf/rte_mbuf_ptype.c b/lib/librte_mbuf/rte_mbuf_ptype.c
index e5c4fae..a450814 100644
--- a/lib/librte_mbuf/rte_mbuf_ptype.c
+++ b/lib/librte_mbuf/rte_mbuf_ptype.c
@@ -89,6 +89,8 @@ const char *rte_get_ptype_tunnel_name(uint32_t ptype)
case RTE_PTYPE_TUNNEL_NVGRE: return "TUNNEL_NVGRE";
case RTE_PTYPE_TUNNEL_GENEVE: return "TUNNEL_GENEVE";
case RTE_PTYPE_TUNNEL_GRENAT: return "TUNNEL_GRENAT";
+   case RTE_PTYPE_TUNNEL_GTPC: return "TUNNEL_GTPC";
+   case RTE_PTYPE_TUNNEL_GTPU: return "TUNNEL_GTPU";
default: return "TUNNEL_UNKNOWN";
}
 }
diff --git a/lib/librte_mbuf/rte_mbuf_ptype.h b/lib/librte_mbuf/rte_mbuf_ptype.h
index acd70bb..eb7cd2c 100644
--- a/lib/librte_mbuf/rte_mbuf_ptype.h
+++ b/lib/librte_mbuf/rte_mbuf_ptype.h
@@ -383,6 +383,30 @@ extern "C" {
  */
 #define RTE_PTYPE_TUNNEL_GRENAT 0x6000
 /**
+ * GTP-C (GPRS Tunnelling Protocol) control tunneling packet type.
+ * Packet format:
+ * <'ether type'=0x0800
+ * | 'version'=4, 'protocol'=17
+ * | 'destination port'=2123>
+ * or,
+ * <'ether type'=0x86DD
+ * | 'version'=6, 'next header'=17
+ * | 'destination port'=2123>
+ */
+#define RTE_PTYPE_TUNNEL_GTPC   0x7000
+/**
+ * GTP-U (GPRS Tunnelling Protocol) user data tunneling packet type.
+ * Packet format:
+ * <'ether type'=0x0800
+ * | 'version'=4, 'protocol'=17
+ * | 'destination port'=2152>
+ * or,
+ * <'ether type'=0x86DD
+ * | 'version'=6, 'next header'=17
+ * | 'destination port'=2152>
+ */
+#define RTE_PTYPE_TUNNEL_GTPU   0x8000
+/**
  * Mask of tunneling packet types.
  */
 #define RTE_PTYPE_TUNNEL_MASK   0xf000
-- 
2.5.5

[dpdk-dev] [PATCH v3 0/8] GPT-C and GTP-U enabling

[Beilei Xing]

This patch set enables RSS/FDIR/cloud filter for GPT-C and GTP-U.
It depends on Kirill's patch:
http://dpdk.org/ml/archives/dev/2017-September/076035.html

v3 changes:
 - Rework implementation to support the new profile.
 - Add GTPC and GTPU tunnel type in software packet type parser.
 - Update ptype info when loading profile.
 - Fix bug of updating pctype info.


v2 changes:
 - Enable RSS/FDIR/cloud filter dinamicly by checking profile
 - Add GTPC and GTPU items to distinguish rule for GTP-C or GTP-U
 - Rework FDIR/cloud filter enabling function
   

Beilei Xing (8):
  net: support GTP in software packet type parser
  net/i40e: update ptype and pctype info
  net/i40e: support RSS for new pctype
  ethdev: add GTP items to support flow API
  net/i40e: finish integration FDIR with generic flow API
  net/i40e: add FDIR support for GTP-C and GTP-U
  net/i40e: add cloud filter parsing function for GTP
  net/i40e: enable cloud filter for GTP-C and GTP-U

 app/test-pmd/cmdline_flow.c |  52 +++
 app/test-pmd/config.c   |   3 +
 doc/guides/prog_guide/rte_flow.rst  |  18 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |   4 +
 drivers/net/i40e/i40e_ethdev.c  | 530 +-
 drivers/net/i40e/i40e_ethdev.h  | 170 -
 drivers/net/i40e/i40e_fdir.c| 570 +++-
 drivers/net/i40e/i40e_flow.c| 486 
 drivers/net/i40e/rte_pmd_i40e.c |   6 +-
 lib/librte_ether/rte_flow.h |  52 +++
 lib/librte_mbuf/rte_mbuf_ptype.c|   2 +
 lib/librte_mbuf/rte_mbuf_ptype.h|  24 ++
 12 files changed, 1791 insertions(+), 126 deletions(-)

-- 
2.5.5

[dpdk-dev] [PATCH v3 3/8] net/i40e: support RSS for new pctype

[Beilei Xing]

Enable RSS for new pctypes after downloading
new profile.

Signed-off-by: Beilei Xing 
---
 drivers/net/i40e/i40e_ethdev.c | 28 
 1 file changed, 28 insertions(+)

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index dcff8cc..a15d994 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -1924,6 +1924,30 @@ i40e_apply_link_speed(struct rte_eth_dev *dev)
return i40e_phy_conf_link(hw, abilities, speed, true);
 }
 
+static void
+i40e_new_pctype_hash_set(struct i40e_pf *pf, bool enable)
+{
+   struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+   uint64_t hena;
+   int i;
+
+   hena = (uint64_t)i40e_read_rx_ctl(hw, I40E_PFQF_HENA(0));
+   hena |= ((uint64_t)i40e_read_rx_ctl(hw, I40E_PFQF_HENA(1))) << 32;
+
+   for (i = 0; i < I40E_PERSONALIZED_MAX; i++) {
+   if (pf->new_pctype[i].valid) {
+   if (enable)
+   hena |= 1ULL << pf->new_pctype[i].pctype;
+   else
+   hena &= ~(1ULL << pf->new_pctype[i].pctype);
+   }
+   }
+
+   i40e_write_rx_ctl(hw, I40E_PFQF_HENA(0), (uint32_t)hena);
+   i40e_write_rx_ctl(hw, I40E_PFQF_HENA(1), (uint32_t)(hena >> 32));
+   I40E_WRITE_FLUSH(hw);
+}
+
 static int
 i40e_dev_start(struct rte_eth_dev *dev)
 {
@@ -2071,6 +2095,8 @@ i40e_dev_start(struct rte_eth_dev *dev)
"please call hierarchy_commit() "
"before starting the port");
 
+   i40e_new_pctype_hash_set(pf, true);
+
return I40E_SUCCESS;
 
 err_up:
@@ -2151,6 +2177,8 @@ i40e_dev_close(struct rte_eth_dev *dev)
uint32_t reg;
int i;
 
+   i40e_new_pctype_hash_set(pf, false);
+
PMD_INIT_FUNC_TRACE();
 
i40e_dev_stop(dev);
-- 
2.5.5

[dpdk-dev] [PATCH v3 2/8] net/i40e: update ptype and pctype info

[Beilei Xing]

Update new packet type and new pctype info when downloading
profile.

Signed-off-by: Beilei Xing 
---
 drivers/net/i40e/i40e_ethdev.c  | 312 
 drivers/net/i40e/i40e_ethdev.h  |  24 
 drivers/net/i40e/rte_pmd_i40e.c |   6 +-
 3 files changed, 341 insertions(+), 1 deletion(-)

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 720f067..dcff8cc 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -65,6 +65,7 @@
 #include "i40e_rxtx.h"
 #include "i40e_pf.h"
 #include "i40e_regs.h"
+#include "rte_pmd_i40e.h"
 
 #define ETH_I40E_FLOATING_VEB_ARG  "enable_floating_veb"
 #define ETH_I40E_FLOATING_VEB_LIST_ARG "floating_veb_list"
@@ -1036,6 +1037,18 @@ i40e_init_fdir_filter_list(struct rte_eth_dev *dev)
return ret;
 }
 
+static void
+i40e_init_customer_pctype(struct i40e_pf *pf)
+{
+   int i;
+
+   for (i = I40E_PERSONALIZED_GTPC; i < I40E_PERSONALIZED_MAX; i++) {
+   pf->new_pctype[i].index = i;
+   pf->new_pctype[i].pctype = I40E_FILTER_PCTYPE_INVALID;
+   pf->new_pctype[i].valid = false;
+   }
+}
+
 static int
 eth_i40e_dev_init(struct rte_eth_dev *dev)
 {
@@ -1301,6 +1314,8 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)
/* initialize Traffic Manager configuration */
i40e_tm_conf_init(dev);
 
+   i40e_init_customer_pctype(pf);
+
ret = i40e_init_ethtype_filter_list(dev);
if (ret < 0)
goto err_init_ethtype_filter_list;
@@ -10893,6 +10908,303 @@ is_i40e_supported(struct rte_eth_dev *dev)
return is_device_supported(dev, &rte_i40e_pmd);
 }
 
+struct i40e_personalized_pctype*
+i40e_find_personalized_pctype(struct i40e_pf *pf, uint8_t index)
+{
+   int i;
+
+   for (i = 0; i < I40E_PERSONALIZED_MAX; i++) {
+   if (pf->new_pctype[i].index == index)
+   return &pf->new_pctype[i];
+   }
+   return NULL;
+}
+
+static int
+i40e_update_personalized_pctype(struct rte_eth_dev *dev, uint8_t *pkg,
+   uint32_t pkg_size, uint32_t proto_num,
+   struct rte_pmd_i40e_proto_info *proto)
+{
+   struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+   uint32_t pctype_num;
+   struct rte_pmd_i40e_ptype_info *pctype;
+   struct i40e_personalized_pctype *new_pctype = NULL;
+   uint8_t proto_id;
+   uint8_t pctype_value;
+   char name[64];
+   uint32_t i, j, n;
+   int ret;
+
+   ret = rte_pmd_i40e_get_ddp_info(pkg, pkg_size,
+   (uint8_t *)&pctype_num, sizeof(pctype_num),
+   RTE_PMD_I40E_PKG_INFO_PCTYPE_NUM);
+   if (ret) {
+   PMD_DRV_LOG(ERR, "Failed to get pctype number");
+   return -1;
+   }
+   if (!pctype_num) {
+   PMD_DRV_LOG(INFO, "No new pctype added");
+   return -1;
+   }
+
+   pctype = rte_zmalloc("new_pctype",
+   pctype_num * sizeof(struct rte_pmd_i40e_ptype_info),
+   0);
+   if (!pctype) {
+   PMD_DRV_LOG(ERR, "Failed to allocate memory");
+   return -1;
+   }
+   /* get information about new pctype list */
+   ret = rte_pmd_i40e_get_ddp_info(pkg, pkg_size,
+   (uint8_t *)pctype, pctype_num,
+   RTE_PMD_I40E_PKG_INFO_PCTYPE_LIST);
+   if (ret) {
+   PMD_DRV_LOG(ERR, "Failed to get pctype list");
+   rte_free(pctype);
+   return -1;
+   }
+
+   /* Update personalized pctype. */
+   for (i = 0; i < pctype_num; i++) {
+   pctype_value = pctype[i].ptype_id;
+   memset(name, 0, sizeof(name));
+   for (j = 0; j < RTE_PMD_I40E_PROTO_NUM; j++) {
+   proto_id = pctype[i].protocols[j];
+   if (proto_id == RTE_PMD_I40E_PROTO_UNUSED)
+   continue;
+   for (n = 0; n < proto_num; n++) {
+   if (proto[n].proto_id != proto_id)
+   continue;
+   strcat(name, proto[n].name);
+   strcat(name, "_");
+   break;
+   }
+   }
+   if (!memcmp(name, "GTPC", sizeof("GTPC") - 1))
+   new_pctype =
+   i40e_find_personalized_pctype(pf,
+ I40E_PERSONALIZED_GTPC);
+   else if (!memcmp(name, "GTPU_IPV4",
+sizeof("GTPU_IPV4") - 1))
+   new_pctype =
+   i40e_find_personalized_pctype(pf,
+  I40E_PERSON

[dpdk-dev] [PATCH v3 4/8] ethdev: add GTP items to support flow API

[Beilei Xing]

This patch adds GTP, GTPC and GTPU items for
generic flow API, and also exposes item fields
through the flow command.

Signed-off-by: Beilei Xing 
---
 app/test-pmd/cmdline_flow.c | 52 +
 app/test-pmd/config.c   |  3 ++
 doc/guides/prog_guide/rte_flow.rst  | 18 ++
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  4 +++
 lib/librte_ether/rte_flow.h | 52 +
 5 files changed, 129 insertions(+)

diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
index a17a004..5e291c8 100644
--- a/app/test-pmd/cmdline_flow.c
+++ b/app/test-pmd/cmdline_flow.c
@@ -171,6 +171,10 @@ enum index {
ITEM_GRE_PROTO,
ITEM_FUZZY,
ITEM_FUZZY_THRESH,
+   ITEM_GTP,
+   ITEM_GTP_TEID,
+   ITEM_GTPC,
+   ITEM_GTPU,
 
/* Validate/create actions. */
ACTIONS,
@@ -451,6 +455,9 @@ static const enum index next_item[] = {
ITEM_MPLS,
ITEM_GRE,
ITEM_FUZZY,
+   ITEM_GTP,
+   ITEM_GTPC,
+   ITEM_GTPU,
ZERO,
 };
 
@@ -588,6 +595,24 @@ static const enum index item_gre[] = {
ZERO,
 };
 
+static const enum index item_gtp[] = {
+   ITEM_GTP_TEID,
+   ITEM_NEXT,
+   ZERO,
+};
+
+static const enum index item_gtpc[] = {
+   ITEM_GTP_TEID,
+   ITEM_NEXT,
+   ZERO,
+};
+
+static const enum index item_gtpu[] = {
+   ITEM_GTP_TEID,
+   ITEM_NEXT,
+   ZERO,
+};
+
 static const enum index next_action[] = {
ACTION_END,
ACTION_VOID,
@@ -1421,6 +1446,33 @@ static const struct token token_list[] = {
.args = ARGS(ARGS_ENTRY(struct rte_flow_item_fuzzy,
thresh)),
},
+   [ITEM_GTP] = {
+   .name = "gtp",
+   .help = "match GTP header",
+   .priv = PRIV_ITEM(GTP, sizeof(struct rte_flow_item_gtp)),
+   .next = NEXT(item_gtp),
+   .call = parse_vc,
+   },
+   [ITEM_GTP_TEID] = {
+   .name = "teid",
+   .help = "tunnel endpoint identifier",
+   .next = NEXT(item_gtp, NEXT_ENTRY(UNSIGNED), item_param),
+   .args = ARGS(ARGS_ENTRY_HTON(struct rte_flow_item_gtp, teid)),
+   },
+   [ITEM_GTPC] = {
+   .name = "gtpc",
+   .help = "match GTP header",
+   .priv = PRIV_ITEM(GTPC, sizeof(struct rte_flow_item_gtp)),
+   .next = NEXT(item_gtpc),
+   .call = parse_vc,
+   },
+   [ITEM_GTPU] = {
+   .name = "gtpu",
+   .help = "match GTP header",
+   .priv = PRIV_ITEM(GTPU, sizeof(struct rte_flow_item_gtp)),
+   .next = NEXT(item_gtpu),
+   .call = parse_vc,
+   },
 
/* Validate/create actions. */
[ACTIONS] = {
diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c
index ca83eef..4063e01 100644
--- a/app/test-pmd/config.c
+++ b/app/test-pmd/config.c
@@ -947,6 +947,9 @@ static const struct {
MK_FLOW_ITEM(MPLS, sizeof(struct rte_flow_item_mpls)),
MK_FLOW_ITEM(GRE, sizeof(struct rte_flow_item_gre)),
MK_FLOW_ITEM(FUZZY, sizeof(struct rte_flow_item_fuzzy)),
+   MK_FLOW_ITEM(GTP, sizeof(struct rte_flow_item_gtp)),
+   MK_FLOW_ITEM(GTPC, sizeof(struct rte_flow_item_gtp)),
+   MK_FLOW_ITEM(GTPU, sizeof(struct rte_flow_item_gtp)),
 };
 
 /** Compute storage space needed by item specification. */
diff --git a/doc/guides/prog_guide/rte_flow.rst 
b/doc/guides/prog_guide/rte_flow.rst
index 662a912..1bc8f19 100644
--- a/doc/guides/prog_guide/rte_flow.rst
+++ b/doc/guides/prog_guide/rte_flow.rst
@@ -955,6 +955,24 @@ Usage example, fuzzy match a TCPv4 packets:
| 4 | END  |
+---+--+
 
+Item: ``GTP``, ``GTPC``, ``GTPU``
+^
+
+Matches a GTP header.
+
+Note: GTP, GTPC and GTPU use the same structure. Since only UDP destination 
port
+is used to distinguish GTP_C (port is 2123) and GTP_U packets (port is 2152),
+GTPC and GTPU item are defined for a user-friendly API when creating GTP-C and
+GTP-U flow.
+
+- ``v_pt_rsv_flags``: version (3b), protocol type (1b), reserved (1b),
+  extension header flag (1b), sequence number flag (1b), N-PDU number
+  flag (1b).
+- ``msg_type``: message type.
+- ``msg_len``: message length.
+- ``teid``: tunnel endpoint identifier.
+- Default ``mask`` matches teid only.
+
 Actions
 ~~~
 
diff --git a/doc/guides/testpmd_app_ug/testpmd_funcs.rst 
b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
index 2ed62f5..8cc2399 100644
--- a/doc/guides/testpmd_app_ug/testpmd_funcs.rst
+++ b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
@@ -2696,6 +2696,10 @@ This section lists supported pattern items and their 
attributes, if any.
 
   - ``thresh {unsigned}``: accuracy threshold.
 
+- ``gtp``, ``gtpc``, ``gtpu``: match GTP header.
+
+  - ``teid {unsigned}``: tunnel endpoi

[dpdk-dev] [PATCH v3 7/8] net/i40e: add cloud filter parsing function for GTP

[Beilei Xing]

This patch adds i40e_flow_parse_gtp_filter parsing
function for GTP-C and GTP-U.

Signed-off-by: Beilei Xing 
---
 drivers/net/i40e/i40e_ethdev.h |   2 +
 drivers/net/i40e/i40e_flow.c   | 151 +
 2 files changed, 153 insertions(+)

diff --git a/drivers/net/i40e/i40e_ethdev.h b/drivers/net/i40e/i40e_ethdev.h
index 928a068..913d9b1 100644
--- a/drivers/net/i40e/i40e_ethdev.h
+++ b/drivers/net/i40e/i40e_ethdev.h
@@ -720,6 +720,8 @@ enum i40e_tunnel_type {
I40E_TUNNEL_TYPE_MPLSoUDP,
I40E_TUNNEL_TYPE_MPLSoGRE,
I40E_TUNNEL_TYPE_QINQ,
+   I40E_TUNNEL_TYPE_GTPC,
+   I40E_TUNNEL_TYPE_GTPU,
I40E_TUNNEL_TYPE_MAX,
 };
 
diff --git a/drivers/net/i40e/i40e_flow.c b/drivers/net/i40e/i40e_flow.c
index b9eaf17..f37ab10 100644
--- a/drivers/net/i40e/i40e_flow.c
+++ b/drivers/net/i40e/i40e_flow.c
@@ -125,6 +125,12 @@ static int i40e_flow_parse_mpls_filter(struct rte_eth_dev 
*dev,
   const struct rte_flow_action actions[],
   struct rte_flow_error *error,
   union i40e_filter_t *filter);
+static int i40e_flow_parse_gtp_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_flow_error *error,
+ union i40e_filter_t *filter);
 static int i40e_flow_destroy_ethertype_filter(struct i40e_pf *pf,
  struct i40e_ethertype_filter *filter);
 static int i40e_flow_destroy_tunnel_filter(struct i40e_pf *pf,
@@ -1808,6 +1814,11 @@ static struct i40e_valid_pattern 
i40e_supported_patterns[] = {
{ pattern_mpls_2, i40e_flow_parse_mpls_filter },
{ pattern_mpls_3, i40e_flow_parse_mpls_filter },
{ pattern_mpls_4, i40e_flow_parse_mpls_filter },
+   /* GTP-C & GTP-U */
+   { pattern_fdir_ipv4_gtpc, i40e_flow_parse_gtp_filter },
+   { pattern_fdir_ipv4_gtpu, i40e_flow_parse_gtp_filter },
+   { pattern_fdir_ipv6_gtpc, i40e_flow_parse_gtp_filter },
+   { pattern_fdir_ipv6_gtpu, i40e_flow_parse_gtp_filter },
/* QINQ */
{ pattern_qinq_1, i40e_flow_parse_qinq_filter },
 };
@@ -3813,6 +3824,146 @@ i40e_flow_parse_mpls_filter(struct rte_eth_dev *dev,
 }
 
 /* 1. Last in item should be NULL as range is not supported.
+ * 2. Supported filter types: GTP TEID.
+ * 3. Mask of fields which need to be matched should be
+ *filled with 1.
+ * 4. Mask of fields which needn't to be matched should be
+ *filled with 0.
+ */
+static int
+i40e_flow_parse_gtp_pattern(struct rte_eth_dev *dev,
+   const struct rte_flow_item *pattern,
+   struct rte_flow_error *error,
+   struct i40e_tunnel_filter_conf *filter)
+{
+   struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+   const struct rte_flow_item *item = pattern;
+   const struct rte_flow_item_gtp *gtp_spec;
+   const struct rte_flow_item_gtp *gtp_mask;
+   enum rte_flow_item_type item_type;
+
+   for (; item->type != RTE_FLOW_ITEM_TYPE_END; item++) {
+   if (item->last) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ITEM,
+  item,
+  "Not support range");
+   return -rte_errno;
+   }
+   item_type = item->type;
+   switch (item_type) {
+   case RTE_FLOW_ITEM_TYPE_ETH:
+   if (item->spec || item->mask) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ITEM,
+  item,
+  "Invalid ETH item");
+   return -rte_errno;
+   }
+   break;
+   case RTE_FLOW_ITEM_TYPE_IPV4:
+   filter->ip_type = I40E_TUNNEL_IPTYPE_IPV4;
+   /* IPv4 is used to describe protocol,
+* spec and mask should be NULL.
+*/
+   if (item->spec || item->mask) {
+   rte_flow_error_set(error, EINVAL,
+  RTE_FLOW_ERROR_TYPE_ITEM,
+  item,
+  "Invalid IPv4 item");
+   return -rte_errno;
+   }
+   break;
+   case RTE_FLOW_ITE

[dpdk-dev] [PATCH v3 5/8] net/i40e: finish integration FDIR with generic flow API

[Beilei Xing]

rte_eth_fdir_* structures are still used in FDIR functions.
This patch adds i40e private FDIR related structures and
functions to finish integration FDIR with generic flow API.

Signed-off-by: Beilei Xing 
---
 drivers/net/i40e/i40e_ethdev.h |  94 +++-
 drivers/net/i40e/i40e_fdir.c   | 490 +++--
 drivers/net/i40e/i40e_flow.c   |  76 +++
 3 files changed, 597 insertions(+), 63 deletions(-)

diff --git a/drivers/net/i40e/i40e_ethdev.h b/drivers/net/i40e/i40e_ethdev.h
index 85ae07b..bec68c4 100644
--- a/drivers/net/i40e/i40e_ethdev.h
+++ b/drivers/net/i40e/i40e_ethdev.h
@@ -460,6 +460,91 @@ struct i40e_vmdq_info {
 #define I40E_FLEX_WORD_MASK(off) (0x80 >> (off))
 #define I40E_FDIR_IPv6_TC_OFFSET   20
 
+/**
+ * A union contains the inputs for all types of flow
+ * Items in flows need to be in big endian
+ */
+union i40e_fdir_flow {
+   struct rte_eth_l2_flow l2_flow;
+   struct rte_eth_udpv4_flow  udp4_flow;
+   struct rte_eth_tcpv4_flow  tcp4_flow;
+   struct rte_eth_sctpv4_flow sctp4_flow;
+   struct rte_eth_ipv4_flow   ip4_flow;
+   struct rte_eth_udpv6_flow  udp6_flow;
+   struct rte_eth_tcpv6_flow  tcp6_flow;
+   struct rte_eth_sctpv6_flow sctp6_flow;
+   struct rte_eth_ipv6_flow   ipv6_flow;
+};
+
+/**
+ * A structure used to contain extend input of flow
+ */
+struct i40e_fdir_flow_ext {
+   uint16_t vlan_tci;
+   uint8_t flexbytes[RTE_ETH_FDIR_MAX_FLEXLEN];
+   /**< It is filled by the flexible payload to match. */
+   uint8_t is_vf;   /**< 1 for VF, 0 for port dev */
+   uint16_t dst_id; /**< VF ID, available when is_vf is 1*/
+};
+
+/**
+ * A structure used to define the input for a flow director filter entry
+ */
+struct i40e_fdir_input {
+   enum i40e_filter_pctype pctype;
+   union i40e_fdir_flow flow;
+   /**< Flow fields to match, dependent on flow_type */
+   struct i40e_fdir_flow_ext flow_ext;
+   /**< Additional fields to match */
+};
+
+/**
+ * Behavior will be taken if FDIR match
+ */
+enum i40e_fdir_behavior {
+   I40E_FDIR_ACCEPT = 0,
+   I40E_FDIR_REJECT,
+   I40E_FDIR_PASSTHRU,
+};
+
+/**
+ * Flow director report status
+ * It defines what will be reported if FDIR entry is matched.
+ */
+enum i40e_fdir_status {
+   I40E_FDIR_NO_REPORT_STATUS = 0, /**< Report nothing. */
+   I40E_FDIR_REPORT_ID,/**< Only report FD ID. */
+   I40E_FDIR_REPORT_ID_FLEX_4, /**< Report FD ID and 4 flex bytes. */
+   I40E_FDIR_REPORT_FLEX_8,/**< Report 8 flex bytes. */
+};
+
+/**
+ * A structure used to define an action when match FDIR packet filter.
+ */
+struct i40e_fdir_action {
+   uint16_t rx_queue;/**< Queue assigned to if FDIR match. */
+   enum i40e_fdir_behavior behavior; /**< Behavior will be taken */
+   enum i40e_fdir_status report_status;  /**< Status report option */
+   /**
+* If report_status is I40E_FDIR_REPORT_ID_FLEX_4 or
+* I40E_FDIR_REPORT_FLEX_8, flex_off specifies where the reported
+* flex bytes start from in flexible payload.
+*/
+   uint8_t flex_off;
+};
+
+/**
+ * A structure used to define the flow director filter entry by filter_ctrl API
+ * It supports RTE_ETH_FILTER_FDIR with RTE_ETH_FILTER_ADD and
+ * RTE_ETH_FILTER_DELETE operations.
+ */
+struct i40e_fdir_filter_conf {
+   uint32_t soft_id;
+   /**< ID, an unique value is required when deal with FDIR entry */
+   struct i40e_fdir_input input;/**< Input set */
+   struct i40e_fdir_action action;  /**< Action taken when match */
+};
+
 /*
  * Structure to store flex pit for flow diretor.
  */
@@ -483,7 +568,7 @@ struct i40e_fdir_flex_mask {
 
 struct i40e_fdir_filter {
TAILQ_ENTRY(i40e_fdir_filter) rules;
-   struct rte_eth_fdir_filter fdir;
+   struct i40e_fdir_filter_conf fdir;
 };
 
 TAILQ_HEAD(i40e_fdir_filter_list, i40e_fdir_filter);
@@ -907,7 +992,7 @@ extern const struct rte_flow_ops i40e_flow_ops;
 
 union i40e_filter_t {
struct rte_eth_ethertype_filter ethertype_filter;
-   struct rte_eth_fdir_filter fdir_filter;
+   struct i40e_fdir_filter_conf fdir_filter;
struct rte_eth_tunnel_filter_conf tunnel_filter;
struct i40e_tunnel_filter_conf consistent_tunnel_filter;
 };
@@ -981,7 +1066,7 @@ i40e_sw_ethertype_filter_lookup(struct i40e_ethertype_rule 
*ethertype_rule,
 int i40e_sw_ethertype_filter_del(struct i40e_pf *pf,
 struct i40e_ethertype_filter_input *input);
 int i40e_sw_fdir_filter_del(struct i40e_pf *pf,
-   struct rte_eth_fdir_input *input);
+   struct i40e_fdir_input *input);
 struct i40e_tunnel_filter *
 i40e_sw_tunnel_filter_lookup(struct i40e_tunnel_rule *tunnel_rule,
 const struct i40e_tunnel_filter_input *input);
@@ -994,6 +1079,9 @@ int i40e_ethertype_filter_set(struct i40e_pf *pf,
 int i40e_add_

[dpdk-dev] [PATCH v3 8/8] net/i40e: enable cloud filter for GTP-C and GTP-U

[Beilei Xing]

GTP-C & GTP-U are not supported by cloud filter due
to limited resource of HW, this patch enables GTP-C
and GTP-U cloud filter by replacing inner_mac and
TUNNEL_KEY.
This configuration will be set when adding GTP-C or
GTP-U filter rules, and it will be invalid only by
NIC core reset.

Signed-off-by: Beilei Xing 
---
 drivers/net/i40e/i40e_ethdev.c | 189 +
 drivers/net/i40e/i40e_ethdev.h |  13 ++-
 drivers/net/i40e/i40e_flow.c   |  12 +--
 3 files changed, 187 insertions(+), 27 deletions(-)

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 8514ba0..c93b106 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -7157,7 +7157,7 @@ i40e_status_code i40e_replace_mpls_l1_filter(struct 
i40e_pf *pf)
/* create L1 filter */
filter_replace.old_filter_type =
I40E_AQC_REPLACE_CLOUD_CMD_INPUT_FV_IMAC;
-   filter_replace.new_filter_type = I40E_AQC_ADD_L1_FILTER_TEID_MPLS;
+   filter_replace.new_filter_type = I40E_AQC_ADD_L1_FILTER_0X11;
filter_replace.tr_bit = 0;
 
/* Prepare the buffer, 3 entries */
@@ -7205,12 +7205,12 @@ i40e_status_code i40e_replace_mpls_cloud_filter(struct 
i40e_pf *pf)
I40E_AQC_MIRROR_CLOUD_FILTER;
filter_replace.old_filter_type = I40E_AQC_ADD_CLOUD_FILTER_IIP;
filter_replace.new_filter_type =
-   I40E_AQC_ADD_CLOUD_FILTER_TEID_MPLSoUDP;
+   I40E_AQC_ADD_CLOUD_FILTER_0X11;
/* Prepare the buffer, 2 entries */
filter_replace_buf.data[0] = I40E_AQC_REPLACE_CLOUD_CMD_INPUT_FV_STAG;
filter_replace_buf.data[0] |=
I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
-   filter_replace_buf.data[4] = I40E_AQC_ADD_L1_FILTER_TEID_MPLS;
+   filter_replace_buf.data[4] = I40E_AQC_ADD_L1_FILTER_0X11;
filter_replace_buf.data[4] |=
I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
status = i40e_aq_replace_cloud_filters(hw, &filter_replace,
@@ -7228,12 +7228,131 @@ i40e_status_code i40e_replace_mpls_cloud_filter(struct 
i40e_pf *pf)
I40E_AQC_MIRROR_CLOUD_FILTER;
filter_replace.old_filter_type = I40E_AQC_ADD_CLOUD_FILTER_IMAC;
filter_replace.new_filter_type =
-   I40E_AQC_ADD_CLOUD_FILTER_TEID_MPLSoGRE;
+   I40E_AQC_ADD_CLOUD_FILTER_0X12;
/* Prepare the buffer, 2 entries */
filter_replace_buf.data[0] = I40E_AQC_REPLACE_CLOUD_CMD_INPUT_FV_STAG;
filter_replace_buf.data[0] |=
I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
-   filter_replace_buf.data[4] = I40E_AQC_ADD_L1_FILTER_TEID_MPLS;
+   filter_replace_buf.data[4] = I40E_AQC_ADD_L1_FILTER_0X11;
+   filter_replace_buf.data[4] |=
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
+
+   status = i40e_aq_replace_cloud_filters(hw, &filter_replace,
+  &filter_replace_buf);
+   return status;
+}
+
+static enum i40e_status_code
+i40e_replace_gtp_l1_filter(struct i40e_pf *pf)
+{
+   struct i40e_aqc_replace_cloud_filters_cmd  filter_replace;
+   struct i40e_aqc_replace_cloud_filters_cmd_buf  filter_replace_buf;
+   struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+   enum i40e_status_code status = I40E_SUCCESS;
+
+   /* For GTP-C */
+   memset(&filter_replace, 0,
+  sizeof(struct i40e_aqc_replace_cloud_filters_cmd));
+   memset(&filter_replace_buf, 0,
+  sizeof(struct i40e_aqc_replace_cloud_filters_cmd_buf));
+   /* create L1 filter */
+   filter_replace.old_filter_type =
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_FV_IMAC;
+   filter_replace.new_filter_type = I40E_AQC_ADD_L1_FILTER_0X12;
+   filter_replace.tr_bit = I40E_AQC_NEW_TR_22 |
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
+   /* Prepare the buffer, 2 entries */
+   filter_replace_buf.data[0] =
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_FV_TEID_WORD0;
+   filter_replace_buf.data[0] |=
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
+   filter_replace_buf.data[2] = 0xFF;
+   filter_replace_buf.data[3] = 0xFF;
+   filter_replace_buf.data[4] =
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_FV_TEID_WORD1;
+   filter_replace_buf.data[4] |=
+   I40E_AQC_REPLACE_CLOUD_CMD_INPUT_VALIDATED;
+   filter_replace_buf.data[6] = 0xFF;
+   filter_replace_buf.data[7] = 0xFF;
+   status = i40e_aq_replace_cloud_filters(hw, &filter_replace,
+  &filter_replace_buf);
+   if (status < 0)
+   return status;
+
+   /* for GTP-U */
+   memset(&filter_replace, 0,
+  sizeof(struct i40e_aqc_replace_cloud_filters_cmd));
+   memset(&filter_replace_buf, 0,
+  sizeof(struct i40e_aqc_replace_cloud_filters_cmd_buf));
+   /* create L1 filter */
+   filter_replace.ol

[dpdk-dev] [PATCH v3 6/8] net/i40e: add FDIR support for GTP-C and GTP-U

[Beilei Xing]

This patch adds FDIR support for GTP-C and GTP-U.

Signed-off-by: Beilei Xing 
---
 drivers/net/i40e/i40e_ethdev.c |   1 +
 drivers/net/i40e/i40e_ethdev.h |  37 ++
 drivers/net/i40e/i40e_fdir.c   | 198 ++--
 drivers/net/i40e/i40e_flow.c   | 253 ++---
 4 files changed, 392 insertions(+), 97 deletions(-)

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index a15d994..8514ba0 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -1047,6 +1047,7 @@ i40e_init_customer_pctype(struct i40e_pf *pf)
pf->new_pctype[i].pctype = I40E_FILTER_PCTYPE_INVALID;
pf->new_pctype[i].valid = false;
}
+   pf->new_pctype_used = false;
 }
 
 static int
diff --git a/drivers/net/i40e/i40e_ethdev.h b/drivers/net/i40e/i40e_ethdev.h
index bec68c4..928a068 100644
--- a/drivers/net/i40e/i40e_ethdev.h
+++ b/drivers/net/i40e/i40e_ethdev.h
@@ -233,6 +233,7 @@ enum i40e_flxpld_layer_idx {
 #define I40E_INSET_TUNNEL_SRC_PORT   0x0008ULL
 #define I40E_INSET_TUNNEL_DST_PORT   0x0010ULL
 #define I40E_INSET_TUNNEL_ID 0x0020ULL
+#define I40E_INSET_GTP_TEID  0x0040ULL
 
 /* bit 48 ~ bit 55 */
 #define I40E_INSET_LAST_ETHER_TYPE 0x0001ULL
@@ -461,6 +462,31 @@ struct i40e_vmdq_info {
 #define I40E_FDIR_IPv6_TC_OFFSET   20
 
 /**
+ * A structure used to define the input for GTP flow
+ */
+struct i40e_gtp_flow {
+   struct rte_eth_udpv4_flow udp; /**< IPv4 UDP fields to match. */
+   uint8_t msg_type;  /**< Message type. */
+   uint32_t teid; /**< TEID in big endian. */
+};
+
+/**
+ * A structure used to define the input for GTP IPV4 flow
+ */
+struct i40e_gtp_ipv4_flow {
+   struct i40e_gtp_flow gtp;
+   struct rte_eth_ipv4_flow ip4;
+};
+
+/**
+ * A structure used to define the input for GTP IPV6 flow
+ */
+struct i40e_gtp_ipv6_flow {
+   struct i40e_gtp_flow gtp;
+   struct rte_eth_ipv6_flow ip6;
+};
+
+/**
  * A union contains the inputs for all types of flow
  * Items in flows need to be in big endian
  */
@@ -474,6 +500,14 @@ union i40e_fdir_flow {
struct rte_eth_tcpv6_flow  tcp6_flow;
struct rte_eth_sctpv6_flow sctp6_flow;
struct rte_eth_ipv6_flow   ipv6_flow;
+   struct i40e_gtp_flow   gtp_flow;
+   struct i40e_gtp_ipv4_flow  gtp_ipv4_flow;
+   struct i40e_gtp_ipv6_flow  gtp_ipv6_flow;
+};
+
+enum i40e_fdir_ip_type {
+   I40E_FDIR_IPTYPE_IPV4,
+   I40E_FDIR_IPTYPE_IPV6,
 };
 
 /**
@@ -485,6 +519,8 @@ struct i40e_fdir_flow_ext {
/**< It is filled by the flexible payload to match. */
uint8_t is_vf;   /**< 1 for VF, 0 for port dev */
uint16_t dst_id; /**< VF ID, available when is_vf is 1*/
+   bool inner_ip;   /**< If there is inner ip */
+   enum i40e_fdir_ip_type iip_type; /**< ip type for inner ip */
 };
 
 /**
@@ -891,6 +927,7 @@ struct i40e_pf {
bool gtp_support; /* 1 - support GTP-C and GTP-U */
/* customer personalized pctype */
struct i40e_personalized_pctype new_pctype[I40E_PERSONALIZED_MAX];
+   bool new_pctype_used; /* Check if new PCTYPE is used for FDIR */
 };
 
 enum pending_msg {
diff --git a/drivers/net/i40e/i40e_fdir.c b/drivers/net/i40e/i40e_fdir.c
index eb2593b..56bd2cc 100644
--- a/drivers/net/i40e/i40e_fdir.c
+++ b/drivers/net/i40e/i40e_fdir.c
@@ -71,6 +71,9 @@
 #define I40E_FDIR_IPv6_DEFAULT_HOP_LIMITS   0xFF
 #define I40E_FDIR_IPv6_PAYLOAD_LEN  380
 #define I40E_FDIR_UDP_DEFAULT_LEN   400
+#define I40E_FDIR_GTP_DEFAULT_LEN   384
+#define I40E_FDIR_INNER_IP_DEFAULT_LEN  384
+#define I40E_FDIR_INNER_IPv6_DEFAULT_LEN344
 
 /* Wait time for fdir filter programming */
 #define I40E_FDIR_MAX_WAIT_US 1
@@ -939,16 +942,33 @@ i40e_fdir_construct_pkt(struct i40e_pf *pf,
return 0;
 }
 
+static struct i40e_personalized_pctype *
+i40e_flow_fdir_check_new_pctype(struct i40e_pf *pf, uint8_t pctype)
+{
+   struct i40e_personalized_pctype *cus_pctype;
+   enum i40e_new_proto i = I40E_PERSONALIZED_GTPC;
+
+   for (; i < I40E_PERSONALIZED_MAX; i++) {
+   cus_pctype = &pf->new_pctype[i];
+   if (pctype == cus_pctype->pctype)
+   return cus_pctype;
+   }
+   return NULL;
+}
+
 static inline int
-i40e_flow_fdir_fill_eth_ip_head(const struct i40e_fdir_input *fdir_input,
+i40e_flow_fdir_fill_eth_ip_head(struct i40e_pf *pf,
+   const struct i40e_fdir_input *fdir_input,
unsigned char *raw_pkt,
bool vlan)
 {
+   struct i40e_personalized_pctype *cus_pctype = NULL;
static uint8_t vlan_frame[] = {0x81, 0, 0, 0};
uint16_t *ether_type;
uint8_t len = 2 * sizeof(struct ether_addr);
struct ipv4_hdr *ip;
  

[dpdk-dev] [PATCH v6 0/5] Rework cfgfile API to enable apps config file support

[Kuba Kozak]

New API for cfgfile library allows to create a cfgfile at runtime, add new
section, add entry in a section, update existing entry and save cfgfile
structure to INI file - opens up the possibility to have applications
dynamically build up a proper DPDK configuration, rather than
having to have a pre-existing one. Due the new API functions, simplification
of load() function was made. One new unit test to TEST app was added. It
contains an example of a large INI file whose parsing requires multiple
reallocation of memory.

---
v6:
  Drop changes from v5 for return value in
- rte_cfgfile_section_num_entries(),
- rte_cfgfile_section_entries(),
- rte_cfgfile_section_entries_by_index()
v5:
Some commit message reword
test_cfgfile.c - remove *.ini file created during realloc test
Add #ifdef RTE_NEXT_ABI for return value change in
- rte_cfgfile_section_num_entries(),
- rte_cfgfile_section_entries(),
- rte_cfgfile_section_entries_by_index()
v4: 
Change members of structure cfgfile:
- struct *sections[] to *sections
- struct *entries[] to *entries
- remove free_sections and free_entries
Rework of existing cfgfile API functions to work with modified
rte_cfgfile struct.
Rework of malloc/realloc implementation due rte_cfgfile struct change,
reducing mulitiple mallocs.
Change return error codes for all library functions (errno.h)
Checkpatch fixes
v3: 
split one patchset into two distinct patchsets:
1. cfgfile library and TEST app changes
2. EAL changes and examples (this patchset depends on cfgfile)
v2:
  lib eal:
Rework of rte_eal_configure(struct rte_cfgfile *cfg, char *prgname).
Now this function load data from cfg structure and did initial
initialization of EAL arguments. Vdev argument are stored in different
subsections eg. DPDK.vdev0, DPDK.vdev1 etc. After execution of this
function it is necessary to call rte_eal_init to complete EAL
initialization. There is no more merging arguments from different
sources (cfg file and command line).
Added non_eal_configure to testpmd application.
Function maintain the same functionality as rte_eal_configure but
for non-eal arguments. 
Added config JSON feature to testpmd last patch from patchset contain
example showing use of .json configuration files.

  lib cfgfile:
Rework of add_section(), add_entry() new implementation
New members allocated_entries/sections, free_entries/sections
in rte_cfgfile structure, change in array of pointers
**sections, **entries instead of *sections[], *entries[]
Add  set_entry() to update/overwrite already existing entry in cfgfile
struct
Add save() function to save on disc cfgfile structure in INI format
Rework of existing load() function  simplifying the code
Add unit test realloc_sections() in TEST app for testing realloc/malloc
of new API functions, add test for save() function

Jacek Piasecki (5):
  cfgfile: remove EAL dependency
  cfgfile: change existing API functions
  cfgfile: add APIs for cfgfile modification
  cfgfile: rework of load function
  test/cfgfile: add new unit test

 lib/Makefile |   3 +-
 lib/librte_cfgfile/Makefile  |   1 +
 lib/librte_cfgfile/rte_cfgfile.c | 419 ++-
 lib/librte_cfgfile/rte_cfgfile.h |  76 
 lib/librte_cfgfile/rte_cfgfile_version.map   |  11 +
 test/test/test_cfgfile.c |  41 +++
 test/test/test_cfgfiles/etc/realloc_sections.ini | 128 +++
 7 files changed, 512 insertions(+), 167 deletions(-)
 create mode 100644 test/test/test_cfgfiles/etc/realloc_sections.ini

-- 
2.7.4

[dpdk-dev] [PATCH v6 1/5] cfgfile: remove EAL dependency

[Kuba Kozak]

From: Jacek Piasecki 

This patch removes the dependency to EAL in cfgfile library.

Signed-off-by: Jacek Piasecki 
Acked-by: Bruce Richardson 
---
 lib/Makefile |  3 +--
 lib/librte_cfgfile/Makefile  |  1 +
 lib/librte_cfgfile/rte_cfgfile.c | 29 +
 3 files changed, 19 insertions(+), 14 deletions(-)

diff --git a/lib/Makefile b/lib/Makefile
index 86caba1..1080a95 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -32,6 +32,7 @@
 include $(RTE_SDK)/mk/rte.vars.mk
 
 DIRS-y += librte_compat
+DIRS-$(CONFIG_RTE_LIBRTE_CFGFILE) += librte_cfgfile
 DIRS-$(CONFIG_RTE_LIBRTE_EAL) += librte_eal
 DIRS-$(CONFIG_RTE_LIBRTE_RING) += librte_ring
 DEPDIRS-librte_ring := librte_eal
@@ -41,8 +42,6 @@ DIRS-$(CONFIG_RTE_LIBRTE_MBUF) += librte_mbuf
 DEPDIRS-librte_mbuf := librte_eal librte_mempool
 DIRS-$(CONFIG_RTE_LIBRTE_TIMER) += librte_timer
 DEPDIRS-librte_timer := librte_eal
-DIRS-$(CONFIG_RTE_LIBRTE_CFGFILE) += librte_cfgfile
-DEPDIRS-librte_cfgfile := librte_eal
 DIRS-$(CONFIG_RTE_LIBRTE_CMDLINE) += librte_cmdline
 DEPDIRS-librte_cmdline := librte_eal
 DIRS-$(CONFIG_RTE_LIBRTE_ETHER) += librte_ether
diff --git a/lib/librte_cfgfile/Makefile b/lib/librte_cfgfile/Makefile
index 755ef11..0bee43e 100644
--- a/lib/librte_cfgfile/Makefile
+++ b/lib/librte_cfgfile/Makefile
@@ -38,6 +38,7 @@ LIB = librte_cfgfile.a
 
 CFLAGS += -O3
 CFLAGS += $(WERROR_FLAGS)
+CFLAGS += -I$(SRCDIR)/../librte_eal/common/include
 
 EXPORT_MAP := rte_cfgfile_version.map
 
diff --git a/lib/librte_cfgfile/rte_cfgfile.c b/lib/librte_cfgfile/rte_cfgfile.c
index 2588093..9dc25cc 100644
--- a/lib/librte_cfgfile/rte_cfgfile.c
+++ b/lib/librte_cfgfile/rte_cfgfile.c
@@ -36,7 +36,6 @@
 #include 
 #include 
 #include 
-#include 
 
 #include "rte_cfgfile.h"
 
@@ -258,19 +257,25 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
 
struct rte_cfgfile_section *sect =
cfg->sections[curr_section];
-   int n;
+
char *split[2] = {NULL};
-   n = rte_strsplit(buffer, sizeof(buffer), split, 2, '=');
-   if (flags & CFG_FLAG_EMPTY_VALUES) {
-   if ((n < 1) || (n > 2)) {
-   printf("Error at line %d - cannot split 
string, n=%d\n",
-  lineno, n);
-   goto error1;
-   }
+   split[0] = buffer;
+   split[1] = memchr(buffer, '=', len);
+
+   /* when delimeter not found */
+   if (split[1] == NULL) {
+   printf("Error at line %d - cannot "
+   "split string\n", lineno);
+   goto error1;
} else {
-   if (n != 2) {
-   printf("Error at line %d - cannot split 
string, n=%d\n",
-  lineno, n);
+   /* when delimeter found */
+   *split[1] = '\0';
+   split[1]++;
+
+   if (!(flags & CFG_FLAG_EMPTY_VALUES) &&
+   (*split[1] == '\0')) {
+   printf("Error at line %d - cannot "
+   "split string\n", lineno);
goto error1;
}
}
-- 
2.7.4

[dpdk-dev] [PATCH v6 3/5] cfgfile: add APIs for cfgfile modification

[Kuba Kozak]

From: Jacek Piasecki 

Extend existing cfgfile library with providing new API functions:

rte_cfgfile_create() - create new cfgfile object
rte_cfgfile_add_section() - add new section to existing cfgfile
object
rte_cfgfile_add_entry() - add new entry to existing cfgfile
object in specified section
rte_cfgfile_set_entry() - update existing entry in cfgfile object
rte_cfgfile_save() - save existing cfgfile object to INI file

This modification allows to create a cfgfile on
runtime and opens up the possibility to have applications
dynamically build up a proper DPDK configuration, rather than having
to have a pre-existing one.

Signed-off-by: Jacek Piasecki 
Acked-by: Bruce Richardson 
---
 lib/librte_cfgfile/rte_cfgfile.c   | 186 +
 lib/librte_cfgfile/rte_cfgfile.h   |  76 
 lib/librte_cfgfile/rte_cfgfile_version.map |  11 ++
 3 files changed, 273 insertions(+)

diff --git a/lib/librte_cfgfile/rte_cfgfile.c b/lib/librte_cfgfile/rte_cfgfile.c
index 39f61e5..ad9bb2e 100644
--- a/lib/librte_cfgfile/rte_cfgfile.c
+++ b/lib/librte_cfgfile/rte_cfgfile.c
@@ -121,6 +121,36 @@ _get_section(struct rte_cfgfile *cfg, const char 
*sectionname)
 }
 
 static int
+_add_entry(struct rte_cfgfile_section *section, const char *entryname,
+   const char *entryvalue)
+{
+   /* resize entry structure if we don't have room for more entries */
+   if (section->num_entries == section->allocated_entries) {
+   struct rte_cfgfile_entry *n_entries = realloc(
+   section->entries,
+   sizeof(struct rte_cfgfile_entry) *
+   ((section->allocated_entries) +
+   CFG_ALLOC_ENTRY_BATCH));
+
+   if (n_entries == NULL)
+   return -ENOMEM;
+
+   section->entries = n_entries;
+   section->allocated_entries += CFG_ALLOC_ENTRY_BATCH;
+   }
+   /* fill up entry fields with key name and value */
+   struct rte_cfgfile_entry *curr_entry =
+   §ion->entries[section->num_entries];
+
+   snprintf(curr_entry->name, sizeof(curr_entry->name), "%s", entryname);
+   snprintf(curr_entry->value,
+   sizeof(curr_entry->value), "%s", entryvalue);
+   section->num_entries++;
+
+   return 0;
+}
+
+static int
 rte_cfgfile_check_params(const struct rte_cfgfile_parameters *params)
 {
unsigned int valid_comment;
@@ -346,6 +376,162 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
return NULL;
 }
 
+struct rte_cfgfile *
+rte_cfgfile_create(int flags)
+{
+   int i;
+   struct rte_cfgfile *cfg = NULL;
+
+   cfg = malloc(sizeof(*cfg));
+
+   if (cfg == NULL)
+   return NULL;
+
+   cfg->flags = flags;
+   cfg->num_sections = 0;
+
+   /* allocate first batch of sections and entries */
+   cfg->sections = malloc(sizeof(struct rte_cfgfile_section) *
+   CFG_ALLOC_SECTION_BATCH);
+
+   if (cfg->sections == NULL)
+   return NULL;
+
+   cfg->allocated_sections = CFG_ALLOC_SECTION_BATCH;
+
+   for (i = 0; i < CFG_ALLOC_SECTION_BATCH; i++) {
+   cfg->sections[i].entries = malloc(sizeof(
+   struct rte_cfgfile_entry) * CFG_ALLOC_ENTRY_BATCH);
+
+   if (cfg->sections[i].entries == NULL)
+   return NULL;
+
+   cfg->sections[i].num_entries = 0;
+   cfg->sections[i].allocated_entries = CFG_ALLOC_ENTRY_BATCH;
+   }
+
+   if (flags & CFG_FLAG_GLOBAL_SECTION)
+   rte_cfgfile_add_section(cfg, "GLOBAL");
+   return cfg;
+}
+
+int
+rte_cfgfile_add_section(struct rte_cfgfile *cfg, const char *sectionname)
+{
+   int i;
+
+   if (cfg == NULL)
+   return -EINVAL;
+
+   if (sectionname == NULL)
+   return -EINVAL;
+
+   /* resize overall struct if we don't have room for more sections */
+   if (cfg->num_sections == cfg->allocated_sections) {
+
+   struct rte_cfgfile_section *n_sections =
+   realloc(cfg->sections,
+   sizeof(struct rte_cfgfile_section) *
+   ((cfg->allocated_sections) +
+   CFG_ALLOC_SECTION_BATCH));
+
+   if (n_sections == NULL)
+   return -ENOMEM;
+
+   for (i = 0; i < CFG_ALLOC_SECTION_BATCH; i++) {
+   n_sections[i + cfg->allocated_sections].num_entries = 0;
+   n_sections[i +
+cfg->allocated_sections].allocated_entries = 0;
+   n_sections[i + cfg->allocated_sections].entries = NULL;
+   }
+   cfg->sections = n_sections;
+   cfg->allocated_sections += CFG_ALLOC

[dpdk-dev] [PATCH v6 2/5] cfgfile: change existing API functions

[Kuba Kozak]

From: Jacek Piasecki 

Change to flat arrays in cfgfile struct force slightly
diffrent data access for most of cfgfile functions.
This patch provides necessary changes in existing API.

Signed-off-by: Jacek Piasecki 
Acked-by: Bruce Richardson 
---
 lib/librte_cfgfile/rte_cfgfile.c | 121 +++
 1 file changed, 58 insertions(+), 63 deletions(-)

diff --git a/lib/librte_cfgfile/rte_cfgfile.c b/lib/librte_cfgfile/rte_cfgfile.c
index 9dc25cc..39f61e5 100644
--- a/lib/librte_cfgfile/rte_cfgfile.c
+++ b/lib/librte_cfgfile/rte_cfgfile.c
@@ -35,6 +35,7 @@
 #include 
 #include 
 #include 
+#include 
 #include 
 
 #include "rte_cfgfile.h"
@@ -42,13 +43,15 @@
 struct rte_cfgfile_section {
char name[CFG_NAME_LEN];
int num_entries;
-   struct rte_cfgfile_entry *entries[0];
+   int allocated_entries;
+   struct rte_cfgfile_entry *entries;
 };
 
 struct rte_cfgfile {
int flags;
int num_sections;
-   struct rte_cfgfile_section *sections[0];
+   int allocated_sections;
+   struct rte_cfgfile_section *sections;
 };
 
 /** when we resize a file structure, how many extra entries
@@ -104,6 +107,19 @@ _strip(char *str, unsigned len)
return newlen;
 }
 
+static struct rte_cfgfile_section *
+_get_section(struct rte_cfgfile *cfg, const char *sectionname)
+{
+   int i;
+
+   for (i = 0; i < cfg->num_sections; i++) {
+   if (strncmp(cfg->sections[i].name, sectionname,
+   sizeof(cfg->sections[0].name)) == 0)
+   return &cfg->sections[i];
+   }
+   return NULL;
+}
+
 static int
 rte_cfgfile_check_params(const struct rte_cfgfile_parameters *params)
 {
@@ -168,17 +184,17 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
if (flags & CFG_FLAG_GLOBAL_SECTION) {
curr_section = 0;
allocated_entries = CFG_ALLOC_ENTRY_BATCH;
-   cfg->sections[curr_section] = malloc(
-   sizeof(*cfg->sections[0]) +
-   sizeof(cfg->sections[0]->entries[0]) *
+   cfg->sections = malloc(
+   sizeof(cfg->sections[0]) +
+   sizeof(cfg->sections[0].entries) *
allocated_entries);
-   if (cfg->sections[curr_section] == NULL) {
+   if (cfg->sections == NULL) {
printf("Error - no memory for global section\n");
goto error1;
}
 
-   snprintf(cfg->sections[curr_section]->name,
-sizeof(cfg->sections[0]->name), "GLOBAL");
+   snprintf(cfg->sections[curr_section].name,
+sizeof(cfg->sections[0].name), "GLOBAL");
}
 
while (fgets(buffer, sizeof(buffer), f) != NULL) {
@@ -213,7 +229,7 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
 
/* close off old section and add start new one */
if (curr_section >= 0)
-   cfg->sections[curr_section]->num_entries =
+   cfg->sections[curr_section].num_entries =
curr_entry + 1;
curr_section++;
 
@@ -235,17 +251,17 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
/* allocate space for new section */
allocated_entries = CFG_ALLOC_ENTRY_BATCH;
curr_entry = -1;
-   cfg->sections[curr_section] = malloc(
-   sizeof(*cfg->sections[0]) +
-   sizeof(cfg->sections[0]->entries[0]) *
+   cfg->sections = malloc(
+   sizeof(cfg->sections[0]) +
+   sizeof(cfg->sections[0].entries) *
allocated_entries);
-   if (cfg->sections[curr_section] == NULL) {
+   if (cfg->sections == NULL) {
printf("Error - no more memory\n");
goto error1;
}
 
-   snprintf(cfg->sections[curr_section]->name,
-   sizeof(cfg->sections[0]->name),
+   snprintf(cfg->sections[curr_section].name,
+   sizeof(cfg->sections[0].name),
"%s", &buffer[1]);
} else {
/* value line */
@@ -255,8 +271,7 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
goto error1;
}
 
-   struct rte_cfgfile_section *sect =
-   cfg->sections[curr_section];
+   struct rte_cfgfile_sec

Re: [dpdk-dev] [PATCH v1 0/10] Policy Based Power Control for Guest

[Thomas Monjalon]

29/08/2017 15:03, Ananyev, Konstantin:
> 
> Hi Dave,
> 
> > This patchset adds the facility for a guest VM to send a policy down to
> > the host that will allow the host to scale up/down cpu frequencies
> > depending on the policy criteria independently of the DPDK app running in
> > the guest.  This differs from the previous vm_power implementation where
> > individual scale up/down requests were send from the guest to the host via
> > virtio-serial.
> > 
> > It's a modification of the vm_power_manager app that runs in the host, and
> > the guest_vm_power_app example app that runs in the guest. This allows the
> > guest to send down a policy to the host via virtio-serial, which then allows
> > the host to scale up/down based on the criteria in the policy, resulting in
> > quicker scale up/down than individual requests coming from the guest.
> > It also means that the DPDK application running in the guest does not need
> > to be modified in any way, it is unaware that it's cores are being scaled
> > up/down, reducing the effort in implementing a power-aware infrastructure.
> > 
> > The usage model is as follows:
> > 1. Set up the VF's and assign to the guest in the usual way.
> > 2. run vm_power_manager on the host, creating a channel to the guest.
> > 3. Start the guest_vm_power_mgr app on the guest, which establishes
> >a virtio-serial channel to the host.
> > 4. Send down the profile for the guest using the "send_profile now" command.
> >There is an example profile hard-coded into guest_vm_power_mgr.
> > 5. Stop the guest_vm_power_mgr and run your normal power-unaware 
> > application.
> > 6. Send traffic into the VFs at varying traffic rates.
> >Observe the frequency change on the host (turbostat -i 1)
> > 
> > The sequence of code changes are as follows:
> > 
> > Firstly, two new API calls are added to the ethdev layer
> > 1. One to convert a VF id to a PF id. In the patchset
> >this id is a MAC address. This is needed so that the host can map the VFs
> >in the profile to PF so in can monitor the traffic on the relevant PF at 
> > the
> >host level.
> > 2. The other function is to read the low-level traffic throughput on the 
> > NIC.
> >Currently this API reads a NIC register for speed, but we are looking at
> >using a more generic way to get these stats, suggestions welcome.
> 
> Why do you need a server (host) to collect RX/TX statistics for VM?
> Such method seems to have a lot of limitations:
> - no clear method to identify to which VM that VF belongs.
> - rely on HW ability to provide such statistics for PF
>   (limited HW support).
> - wouldn't work if PF is not controlled  by the same DPDK app. 
> Why not to make  it client(VM) responsibility to collect that statistics and
> periodically send it to the server? 
> Then server just will have to process that data and make decision.

Any progress Dave?

You have another series "turbo boost API". Does it depends on this one?

Re: [dpdk-dev] [PATCH v1 01/10] net/i40e: add API to convert VF Id to PF Id

[Thomas Monjalon]

25/08/2017 18:02, David Hunt:
> 
> +static inline uint64_t
> +vfid_to_pfid_direct(uint8_t port_id, uint64_t vfid)
> +{
> +   struct rte_eth_dev *dev = &rte_eth_devices[port_id];
> +   uint64_t pfid  = (*dev->dev_ops->vfid_to_pfid)(dev, vfid);
> +   return pfid;
> +}

I would like to comment this API but there is no associated doxygen.

If the application is aware of the VFs, it probably already knows
how PF and VF are associated.

Until now, the functions to control VF from PF are driver-specifics.

[dpdk-dev] [PATCH v6 4/5] cfgfile: rework of load function

[Kuba Kozak]

From: Jacek Piasecki 

New functions added to cfgfile library make it possible
to significantly simplify the code of rte_cfgfile_load_with_params()

This patch shows the new body of this function.

Signed-off-by: Jacek Piasecki 
Acked-by: Bruce Richardson 
---
 lib/librte_cfgfile/rte_cfgfile.c | 157 ---
 1 file changed, 30 insertions(+), 127 deletions(-)

diff --git a/lib/librte_cfgfile/rte_cfgfile.c b/lib/librte_cfgfile/rte_cfgfile.c
index ad9bb2e..124aef5 100644
--- a/lib/librte_cfgfile/rte_cfgfile.c
+++ b/lib/librte_cfgfile/rte_cfgfile.c
@@ -189,10 +189,6 @@ struct rte_cfgfile *
 rte_cfgfile_load_with_params(const char *filename, int flags,
 const struct rte_cfgfile_parameters *params)
 {
-   int allocated_sections = CFG_ALLOC_SECTION_BATCH;
-   int allocated_entries = 0;
-   int curr_section = -1;
-   int curr_entry = -1;
char buffer[CFG_NAME_LEN + CFG_VALUE_LEN + 4] = {0};
int lineno = 0;
struct rte_cfgfile *cfg = NULL;
@@ -204,28 +200,7 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
if (f == NULL)
return NULL;
 
-   cfg = malloc(sizeof(*cfg) + sizeof(cfg->sections[0]) *
-   allocated_sections);
-   if (cfg == NULL)
-   goto error2;
-
-   memset(cfg->sections, 0, sizeof(cfg->sections[0]) * allocated_sections);
-
-   if (flags & CFG_FLAG_GLOBAL_SECTION) {
-   curr_section = 0;
-   allocated_entries = CFG_ALLOC_ENTRY_BATCH;
-   cfg->sections = malloc(
-   sizeof(cfg->sections[0]) +
-   sizeof(cfg->sections[0].entries) *
-   allocated_entries);
-   if (cfg->sections == NULL) {
-   printf("Error - no memory for global section\n");
-   goto error1;
-   }
-
-   snprintf(cfg->sections[curr_section].name,
-sizeof(cfg->sections[0].name), "GLOBAL");
-   }
+   cfg = rte_cfgfile_create(flags);
 
while (fgets(buffer, sizeof(buffer), f) != NULL) {
char *pos = NULL;
@@ -236,13 +211,15 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
"Check if line too long\n", lineno);
goto error1;
}
+   /* skip parsing if comment character found */
pos = memchr(buffer, params->comment_character, len);
-   if (pos != NULL) {
+   if (pos != NULL && (*(pos-1) != '\\')) {
*pos = '\0';
len = pos -  buffer;
}
 
len = _strip(buffer, len);
+   /* skip lines without useful content */
if (buffer[0] != '[' && memchr(buffer, '=', len) == NULL)
continue;
 
@@ -250,128 +227,54 @@ rte_cfgfile_load_with_params(const char *filename, int 
flags,
/* section heading line */
char *end = memchr(buffer, ']', len);
if (end == NULL) {
-   printf("Error line %d - no terminating '['"
+   printf("Error line %d - no terminating ']'"
"character found\n", lineno);
goto error1;
}
*end = '\0';
_strip(&buffer[1], end - &buffer[1]);
 
-   /* close off old section and add start new one */
-   if (curr_section >= 0)
-   cfg->sections[curr_section].num_entries =
-   curr_entry + 1;
-   curr_section++;
-
-   /* resize overall struct if we don't have room for more
-   sections */
-   if (curr_section == allocated_sections) {
-   allocated_sections += CFG_ALLOC_SECTION_BATCH;
-   struct rte_cfgfile *n_cfg = realloc(cfg,
-   sizeof(*cfg) + sizeof(cfg->sections[0])
-   * allocated_sections);
-   if (n_cfg == NULL) {
-   curr_section--;
-   printf("Error - no more memory\n");
-   goto error1;
-   }
-   cfg = n_cfg;
-   }
-
-   /* allocate space for new section */
-   allocated_entries = CFG_ALLOC_ENTRY_BATCH;
-   curr_entry = -1;
-   cfg->sections = malloc(
-   sizeof(cfg->sections[0]) +
- 

[dpdk-dev] [PATCH v6 5/5] test/cfgfile: add new unit test

[Kuba Kozak]

From: Jacek Piasecki 

Load huge realloc_sections.ini file to check malloc/realloc
ability of cfgfile library.

Signed-off-by: Jacek Piasecki 
Acked-by: Bruce Richardson 
---
 test/test/test_cfgfile.c |  41 
 test/test/test_cfgfiles/etc/realloc_sections.ini | 128 +++
 2 files changed, 169 insertions(+)
 create mode 100644 test/test/test_cfgfiles/etc/realloc_sections.ini

diff --git a/test/test/test_cfgfile.c b/test/test/test_cfgfile.c
index 4cc9b14..25e8539 100644
--- a/test/test/test_cfgfile.c
+++ b/test/test/test_cfgfile.c
@@ -111,6 +111,7 @@ _test_cfgfile_sample(struct rte_cfgfile *cfgfile)
return 0;
 }
 
+
 static int
 test_cfgfile_sample1(void)
 {
@@ -154,6 +155,43 @@ test_cfgfile_sample2(void)
 }
 
 static int
+test_cfgfile_realloc_sections(void)
+{
+   struct rte_cfgfile *cfgfile;
+   int ret;
+   const char *value;
+
+   cfgfile = rte_cfgfile_load(CFG_FILES_ETC "/realloc_sections.ini", 0);
+   TEST_ASSERT_NOT_NULL(cfgfile, "Failed to load config file");
+
+   ret = rte_cfgfile_num_sections(cfgfile, NULL, 0);
+   TEST_ASSERT(ret == 9, "Unexpected number of sections: %d", ret);
+
+   ret = rte_cfgfile_has_section(cfgfile, "section9");
+   TEST_ASSERT(ret, "section9 missing");
+
+   ret = rte_cfgfile_section_num_entries(cfgfile, "section3");
+   TEST_ASSERT(ret == 21,
+   "section3 unexpected number of entries: %d", ret);
+
+   ret = rte_cfgfile_section_num_entries(cfgfile, "section9");
+   TEST_ASSERT(ret == 8, "section9 unexpected number of entries: %d", ret);
+
+   value = rte_cfgfile_get_entry(cfgfile, "section9", "key8");
+   TEST_ASSERT(strcmp("value8_section9", value) == 0,
+   "key unexpected value: %s", value);
+
+   ret = rte_cfgfile_save(cfgfile, "/tmp/cfgfile_save.ini");
+   TEST_ASSERT_SUCCESS(ret, "Failed to save *.ini file");
+   remove("/tmp/cfgfile_save.ini");
+
+   ret = rte_cfgfile_close(cfgfile);
+   TEST_ASSERT_SUCCESS(ret, "Failed to close cfgfile");
+
+   return 0;
+}
+
+static int
 test_cfgfile_invalid_section_header(void)
 {
struct rte_cfgfile *cfgfile;
@@ -292,6 +330,9 @@ test_cfgfile(void)
if (test_cfgfile_sample2())
return -1;
 
+   if (test_cfgfile_realloc_sections())
+   return -1;
+
if (test_cfgfile_invalid_section_header())
return -1;
 
diff --git a/test/test/test_cfgfiles/etc/realloc_sections.ini 
b/test/test/test_cfgfiles/etc/realloc_sections.ini
new file mode 100644
index 000..e653e40
--- /dev/null
+++ b/test/test/test_cfgfiles/etc/realloc_sections.ini
@@ -0,0 +1,128 @@
+[section1]
+key1=value1_section1
+key2=value2_section1
+key3=value3_section1
+key4=value4_section1
+key5=value5_section1
+key6=value6_section1
+key7=value7_section1
+key8=value8_section1
+key9=value9_section1
+key10=value10_section1
+key11=value11_section1
+key12=value12_section1
+key13=value13_section1
+key14=value14_section1
+key15=value15_section1
+key16=value16_section1
+key17=value17_section1
+key18=value18_section1
+key19=value19_section1
+key20=value20_section1
+key21=value21_section1
+
+[section2]
+key1=value1_section2
+key2=value2_section2
+key3=value3_section2
+key4=value4_section2
+key5=value5_section2
+key6=value6_section2
+key7=value7_section2
+key8=value8_section2
+key9=value9_section2
+key10=value10_section2
+key11=value11_section2
+key12=value12_section2
+key13=value13_section2
+key14=value14_section2
+key15=value15_section2
+key16=value16_section2
+key17=value17_section2
+key18=value18_section2
+key19=value19_section2
+key20=value20_section2
+key21=value21_section2
+
+[section3]
+key1=value1_section3
+key2=value2_section3
+key3=value3_section3
+key4=value4_section3
+key5=value5_section3
+key6=value6_section3
+key7=value7_section3
+key8=value8_section3
+key9=value9_section3
+key10=value10_section3
+key11=value11_section3
+key12=value12_section3
+key13=value13_section3
+key14=value14_section3
+key15=value15_section3
+key16=value16_section3
+key17=value17_section3
+key18=value18_section3
+key19=value19_section3
+key20=value20_section3
+key21=value21_section3
+
+[section4]
+key1=value1_section4
+key2=value2_section4
+key3=value3_section4
+key4=value4_section4
+key5=value5_section4
+key6=value6_section4
+key7=value7_section4
+key8=value8_section4
+
+[section5]
+key1=value1_section5
+key2=value2_section5
+key3=value3_section5
+key4=value4_section5
+key5=value5_section5
+key6=value6_section5
+key7=value7_section5
+key8=value8_section5
+
+[section6]
+key1=value1_section6
+key2=value2_section6
+key3=value3_section6
+key4=value4_section6
+key5=value5_section6
+key6=value6_section6
+key7=value7_section6
+key8=value8_section6
+
+[section7]
+key1=value1_section7
+key2=value2_section7
+key3=value3_section7
+key4=value4_section7
+key5=value5_section7
+key6=value6_section7
+key7=value7_section7
+key8=value8_section7
+
+[section8]
+key1=value1_secti

Re: [dpdk-dev] [PATCH v5] ether: add support for vtune task tracing

[Thomas Monjalon]

24/07/2017 11:27, Jerin Jacob:
> Documentation changes could move to different patch for better review.

Sorry for late reply,
I think it is better to have doc with code in the same patch.
I will merge them when applying.

Re: [dpdk-dev] [PATCH v1 0/10] Policy Based Power Control for Guest

[Hunt, David]

On 22/9/2017 10:51 AM, Thomas Monjalon wrote:

29/08/2017 15:03, Ananyev, Konstantin:

Hi Dave,


This patchset adds the facility for a guest VM to send a policy down to
the host that will allow the host to scale up/down cpu frequencies
depending on the policy criteria independently of the DPDK app running in
the guest.  This differs from the previous vm_power implementation where
individual scale up/down requests were send from the guest to the host via
virtio-serial.

It's a modification of the vm_power_manager app that runs in the host, and
the guest_vm_power_app example app that runs in the guest. This allows the
guest to send down a policy to the host via virtio-serial, which then allows
the host to scale up/down based on the criteria in the policy, resulting in
quicker scale up/down than individual requests coming from the guest.
It also means that the DPDK application running in the guest does not need
to be modified in any way, it is unaware that it's cores are being scaled
up/down, reducing the effort in implementing a power-aware infrastructure.

The usage model is as follows:
1. Set up the VF's and assign to the guest in the usual way.
2. run vm_power_manager on the host, creating a channel to the guest.
3. Start the guest_vm_power_mgr app on the guest, which establishes
a virtio-serial channel to the host.
4. Send down the profile for the guest using the "send_profile now" command.
There is an example profile hard-coded into guest_vm_power_mgr.
5. Stop the guest_vm_power_mgr and run your normal power-unaware application.
6. Send traffic into the VFs at varying traffic rates.
Observe the frequency change on the host (turbostat -i 1)

The sequence of code changes are as follows:

Firstly, two new API calls are added to the ethdev layer
1. One to convert a VF id to a PF id. In the patchset
this id is a MAC address. This is needed so that the host can map the VFs
in the profile to PF so in can monitor the traffic on the relevant PF at the
host level.
2. The other function is to read the low-level traffic throughput on the NIC.
Currently this API reads a NIC register for speed, but we are looking at
using a more generic way to get these stats, suggestions welcome.

Why do you need a server (host) to collect RX/TX statistics for VM?
Such method seems to have a lot of limitations:
- no clear method to identify to which VM that VF belongs.
- rely on HW ability to provide such statistics for PF
   (limited HW support).
- wouldn't work if PF is not controlled  by the same DPDK app.
Why not to make  it client(VM) responsibility to collect that statistics and
periodically send it to the server?
Then server just will have to process that data and make decision.

Any progress Dave?

You have another series "turbo boost API". Does it depends on this one?


Hi Thomas,

We're still working on updates based on Konstantin's feedback above, and 
hope to have a new patch set submitted to the mailing list early next 
week. This will remove the ethdev layer changes, and uses pre-existing 
stats-api.


In relation to the Turbo patch, they are still independent, but when we 
have the next revision of the Policy patch submitted, I'll do a new 
version of the Turbo patch so that it can be applied on top of the 
policy patch.


Regards,
Dave.

Re: [dpdk-dev] [PATCH v2] app/testpmd: fix stats period can't quit normally in container

[Jianbo Liu]

On 22 September 2017 at 14:21, Phil Yang  wrote:
> In container, the process cannot be terminated by SIGINT/SIGTERM when
> execute with 'stats-period' option.
> Fixed by adding a flag to exit stats period loop after received either
> SIGINT or SIGTERM.
>
> Signed-off-by: Phil Yang 
> ---
>  app/test-pmd/testpmd.c | 13 -
>  1 file changed, 12 insertions(+), 1 deletion(-)
>
> diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
> index e097ee0..3ccc9ba 100644
> --- a/app/test-pmd/testpmd.c
> +++ b/app/test-pmd/testpmd.c
> @@ -183,6 +183,13 @@ uint16_t mbuf_data_size = DEFAULT_MBUF_DATA_SIZE; /**< 
> Mbuf data space size. */
>  uint32_t param_total_num_mbufs = 0;  /**< number of mbufs in all pools - if
>* specified on command-line. */
>  uint16_t stats_period; /**< Period to show statistics (disabled by default) 
> */
> +
> +/*
> + * In container, it cannot terminate the process which running with 
> 'stats-period'
> + * option. Set flag to exit stats period loop after received SIGINT/SIGTERM.
> + */
> +uint8_t f_quit;
> +
>  /*
>   * Configuration of packet segments used by the "txonly" processing engine.
>   */
> @@ -2318,6 +2325,8 @@ signal_handler(int signum)
> rte_latencystats_uninit();
>  #endif
> force_quit();
> +   /* Set flag to indicate the force termination. */
> +   f_quit = 1;
> /* exit with the expected status */
> signal(signum, SIG_DFL);
> kill(getpid(), signum);
> @@ -2435,6 +2444,8 @@ main(int argc, char** argv)
> char c;
> int rc;
>
> +   f_quit = 0;
> +
> printf("No commandline core given, start packet 
> forwarding\n");
> start_packet_forwarding(tx_first);
> if (stats_period != 0) {
> @@ -2444,7 +2455,7 @@ main(int argc, char** argv)
> /* Convert to number of cycles */
> timer_period = stats_period * rte_get_timer_hz();
>
> -   while (1) {
> +   while (f_quit == 0) {
> cur_time = rte_get_timer_cycles();
> diff_time += cur_time - prev_time;
>
> --
> 2.7.4
>

Acked-by: Jianbo Liu 

Re: [dpdk-dev] [PATCH] doc: update failsafe feature list

[Ferruh Yigit]

On 9/19/2017 12:39 PM, Matan Azrad wrote:
> Hi Ferruh
> 
>> -Original Message-
>> From: Ferruh Yigit [mailto:ferruh.yi...@intel.com]
>> Sent: Tuesday, September 19, 2017 2:00 PM
>> To: Matan Azrad ; Gaetan Rivet
>> 
>> Cc: dev@dpdk.org; john.mcnam...@intel.com
>> Subject: Re: [dpdk-dev] [PATCH] doc: update failsafe feature list
>>
>> On 9/19/2017 11:04 AM, Matan Azrad wrote:
>>>
>>> Hi Ferruh
>>>
 -Original Message-
 From: Ferruh Yigit [mailto:ferruh.yi...@intel.com]
 Sent: Tuesday, September 19, 2017 12:27 PM
 To: Matan Azrad ; Gaetan Rivet
 
 Cc: dev@dpdk.org
 Subject: Re: [dpdk-dev] [PATCH] doc: update failsafe feature list

 On 9/14/2017 4:32 PM, Matan Azrad wrote:
> Add supported failsafe features to feature list.
> Remove stats per queue feature from failsafe feature list since
> queue_stats_mapping_set dev op has not implemented yet.
>
> Signed-off-by: Matan Azrad 
> ---
>  doc/guides/nics/features/failsafe.ini | 15 ++-
>  1 file changed, 14 insertions(+), 1 deletion(-)
>
> diff --git a/doc/guides/nics/features/failsafe.ini
> b/doc/guides/nics/features/failsafe.ini
> index a42e344..9f48455 100644
> --- a/doc/guides/nics/features/failsafe.ini
> +++ b/doc/guides/nics/features/failsafe.ini
> @@ -4,20 +4,33 @@
>  ; Refer to default.ini for the full list of available PMD features.
>  ;
>  [Features]
> +Speed capabilities   = Y
>  Link status  = Y
>  Link status event= Y
>  MTU update   = Y
>  Jumbo frame  = Y
> +Scattered Rx = Y
> +LRO  = Y
> +TSO  = Y
>  Promiscuous mode = Y
>  Allmulticast mode= Y
>  Unicast MAC filter   = Y
>  Multicast MAC filter = Y
>  VLAN filter  = Y
> +Ethertype filter = Y
> +N-tuple filter   = Y
> +SYN filter   = Y
> +Tunnel filter= Y
> +Flexible filter  = Y
> +Hash filter  = Y
> +Flow director= Y
>  Flow control = Y
>  Flow API = Y
> +QinQ offload = Y
> +L3 checksum offload  = Y
> +L4 checksum offload  = Y
>  Packet type parsing  = Y
>  Basic stats  = Y
> -Stats per queue  = Y
>  ARMv7= Y
>  ARMv8= Y
>  Power8   = Y

 I am not sure if claiming support for these features is correct.
 Failsafe itself doesn't provide these features, but relies underlying
 hardware which we don't really know what they supports or not in this
>> stage.

>>>
>>> Don't you think that almost all failsafe features rely underlying hardware 
>>> or
>> sub PMDs?
>>
>> You are right, perhaps we should remove all. This is helpful to show what
>> device features are supported. For failsafe, is this information useful?
>>
> Since there are features that failsafe cannot support without any sub PMD 
> dependences
> (for example "Stats per queue") it is useful.

Sorry, I missed your point.

Device feature list documentation is good for:
- End user can easily see what to expect from a device/driver.
- To trace what features implemented for a device.
- To find out which device has a specific desired feature.

For failsafe, it is a virtual overlay device on other physical devices.

The supported architectures and provided documents features can be
useful. But why/how NIC related features can be useful since all they
are coming form underlay devices?

> 
>>>
 OK for dropping "Stats per queue"

>
>>>
> 

Re: [dpdk-dev] [PATCH] hash: fix incorrect eviction counter

[De Lara Guarch, Pablo]

> -Original Message-
> From: Richardson, Bruce
> Sent: Friday, September 22, 2017 9:36 AM
> To: De Lara Guarch, Pablo 
> Cc: dev@dpdk.org; sta...@dpdk.org
> Subject: Re: [PATCH] hash: fix incorrect eviction counter
> 
> On Thu, Sep 21, 2017 at 01:46:46PM +0100, Pablo de Lara wrote:
> > When adding a new entry in a hash table, there is a maximum number of
> > evictions that can be performed. When the counter of these evictions
> > reaches this maximum, the entry cannot be added, as it is considered
> > that the algorithm has encountered an infinite loop.
> >
> > The problem with the current implementation, is that this counter was
> > declared as a static variable.
> > If there are multiple threads adding entries in the same table or in
> > different tables, they should access different counters, one per core
> > and per table.
> >
> > Therefore, an array of counter has been added to the hash table
> > structure.
> >
> > Fixes: 243e93a5046f ("hash: fix unlimited cuckoo path")
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: Pablo de Lara 
> > ---
> Since you appear to be passing this counter through the different function
> calls as parameter, and it gets reset to zero at the start, do you even need
> an array. Can you not just have a non-static local variable in the function?
> 
> /Bruce

Right, I clearly overthought here... Will send a v2 with your much simpler 
approach.

Thanks!
Pablo

Re: [dpdk-dev] [PATCH v7] ether: add support for vtune task tracing

[Thomas Monjalon]

08/09/2017 14:57, ilia.kura...@intel.com:
> + /* See rte_ethdev_profile.h to find comments on code below. */
> + __rte_eth_profile_rx_init(port_id, dev);

We can find easily where the function is defined.
Please replace this comment by something like that:
/* Initialize Rx profiling if enabled at compilation time. */

Why not checking an error code for this initialization?
If we enable it, we expect the application to abort if
the profiling does not work.


> --- /dev/null
> +++ b/lib/librte_ether/rte_ethdev_profile.h

This file is internal so it should not appear in doxygen index.
Please remove rte_ prefix from the filename to blacklist it from
doxygen point of view.
You can apply the same change to the .c file for consistency.

Re: [dpdk-dev] [PATCH v3 1/7] member: implement main API

[Thomas Monjalon]

06/09/2017 01:59, Yipeng Wang:
> --- a/lib/librte_eal/common/eal_common_log.c
> +++ b/lib/librte_eal/common/eal_common_log.c
> @@ -279,6 +279,7 @@ static const struct logtype logtype_strings[] = {
> {RTE_LOGTYPE_CRYPTODEV,  "cryptodev"},
> {RTE_LOGTYPE_EFD,"efd"},
> {RTE_LOGTYPE_EVENTDEV,   "eventdev"},
> +   {RTE_LOGTYPE_MEMBER, "member"},
> {RTE_LOGTYPE_USER1,  "user1"},
> {RTE_LOGTYPE_USER2,  "user2"},
> {RTE_LOGTYPE_USER3,  "user3"},

The static logtypes should be removed.
Please use the new dynamic log types.

[dpdk-dev] DPDK Community Survey 2017

[Jain, Deepak K]

Hi All,

As part of our ongoing efforts to improve DPDK, we'd like to hear your feedback!

We have created a number of DPDK-related questions here.

https://www.surveymonkey.com/r/DPDK_Community_Survey_2017

and want to hear your views!!

The survey will close at midnight GMT on Sunday October 1st, 2017.

Thanks in advance for your feedback - the more responses we get the more data 
we have to drive further features, improvements, etc.

So please respond!!!

Regards
Deepak

Re: [dpdk-dev] [PATCH v3 5/7] member: enable the library

[Thomas Monjalon]

06/09/2017 01:59, Yipeng Wang:
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -734,6 +734,13 @@ F: test/test/test_meter.c
>  F: examples/qos_meter/
>  F: doc/guides/sample_app_ug/qos_metering.rst
>  
> +Membership - EXPERIMENTAL
> +M: Yipeng Wang 
> +M: Sameh Gobriel 
> +F: lib/librte_member/
> +F: doc/guides/prog_guide/member_lib.rst
> +F: test/test/test_member*

I know it is not obvious but this section is sorted in alphabetical order.
Please move between Hashes and LPM. Thanks

[dpdk-dev] [PATCH] ethdev: revert use device name from device structure

[Ferruh Yigit]

This reverts commit a1e7c17555e8f77d520ba5f06ed26c00e77a2bd1.

Original commit assumes there is 1:1 mapping between physical device and
ethdev port, so that device name can be used per port instead of ethdev
name field.

But one physical device may have multiple ethdev ports and each port
needs its own unique name.

One issue reported here:
http://dpdk.org/ml/archives/users/2017-September/002484.html

So reverting back the commit to continue using ethdev name field per
port.

Reported-by: Adrien Mazarguil 
Signed-off-by: Ferruh Yigit 
---
 lib/librte_ether/rte_ethdev.c | 25 ++---
 1 file changed, 10 insertions(+), 15 deletions(-)

diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index 6016fd9be..c944b9fb1 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -177,11 +177,9 @@ rte_eth_dev_allocated(const char *name)
unsigned i;
 
for (i = 0; i < RTE_MAX_ETHPORTS; i++) {
-   if (rte_eth_devices[i].state == RTE_ETH_DEV_ATTACHED &&
-   rte_eth_devices[i].device) {
-   if (!strcmp(rte_eth_devices[i].device->name, name))
-   return &rte_eth_devices[i];
-   }
+   if ((rte_eth_devices[i].state == RTE_ETH_DEV_ATTACHED) &&
+   strcmp(rte_eth_devices[i].data->name, name) == 0)
+   return &rte_eth_devices[i];
}
return NULL;
 }
@@ -318,7 +316,7 @@ rte_eth_dev_count(void)
 int
 rte_eth_dev_get_name_by_port(uint8_t port_id, char *name)
 {
-   const char *tmp;
+   char *tmp;
 
RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -EINVAL);
 
@@ -329,7 +327,7 @@ rte_eth_dev_get_name_by_port(uint8_t port_id, char *name)
 
/* shouldn't check 'rte_eth_devices[i].data',
 * because it might be overwritten by VDEV PMD */
-   tmp = rte_eth_devices[port_id].device->name;
+   tmp = rte_eth_dev_data[port_id].name;
strcpy(name, tmp);
return 0;
 }
@@ -337,7 +335,6 @@ rte_eth_dev_get_name_by_port(uint8_t port_id, char *name)
 int
 rte_eth_dev_get_port_by_name(const char *name, uint8_t *port_id)
 {
-   int ret;
int i;
 
if (name == NULL) {
@@ -346,13 +343,11 @@ rte_eth_dev_get_port_by_name(const char *name, uint8_t 
*port_id)
}
 
RTE_ETH_FOREACH_DEV(i) {
-   if (!rte_eth_devices[i].device)
-   continue;
+   if (!strncmp(name,
+   rte_eth_dev_data[i].name, strlen(name))) {
 
-   ret = strncmp(name, rte_eth_devices[i].device->name,
-   strlen(name));
-   if (ret == 0) {
*port_id = i;
+
return 0;
}
}
@@ -435,8 +430,8 @@ rte_eth_dev_detach(uint8_t port_id, char *name)
if (rte_eth_dev_is_detachable(port_id))
goto err;
 
-   snprintf(name, RTE_DEV_NAME_MAX_LEN, "%s",
-rte_eth_devices[port_id].device->name);
+   snprintf(name, sizeof(rte_eth_devices[port_id].data->name),
+"%s", rte_eth_devices[port_id].data->name);
 
ret = rte_eal_dev_detach(rte_eth_devices[port_id].device);
if (ret < 0)
-- 
2.13.5

Re: [dpdk-dev] [PATCH 06/11] ethdev: extend ethdev to support security APIs

[Radu Nicolau]

Hi


On 9/17/2017 2:45 PM, Shahaf Shuler wrote:

Hi Declan,

Thursday, September 14, 2017 11:27 AM, Akhil Goyal:

From: Declan Doherty 

rte_flow_action type and ethdev updated to support rte_security sessions
for crypto offload to ethernet device.

Signed-off-by: Boris Pismenny 
Signed-off-by: Aviad Yehezkel 
Signed-off-by: Radu Nicolau 
Signed-off-by: Declan Doherty 
---
  lib/librte_ether/rte_ethdev.c   | 11 +++
  lib/librte_ether/rte_ethdev.h   | 22 --
  lib/librte_ether/rte_ethdev_version.map |  7 +++
  3 files changed, 38 insertions(+), 2 deletions(-)

diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index 0597641..f51c5a5 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -302,6 +302,17 @@ rte_eth_dev_socket_id(uint8_t port_id)
return rte_eth_devices[port_id].data->numa_node;
  }

+uint16_t
+rte_eth_dev_get_sec_id(uint8_t port_id) {
+   RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -1);
+
+   if (rte_eth_devices[port_id].data->dev_flags &
RTE_ETH_DEV_SECURITY)
+   return rte_eth_devices[port_id].data->sec_id;
+
+   return -1;
+}
+
  uint8_t
  rte_eth_dev_count(void)
  {
diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index 0adf327..262ba47 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -180,6 +180,8 @@ extern "C" {
  #include 
  #include 
  #include 
+#include 
+
  #include "rte_ether.h"
  #include "rte_eth_ctrl.h"
  #include "rte_dev_info.h"
@@ -357,7 +359,8 @@ struct rte_eth_rxmode {
jumbo_frame  : 1, /**< Jumbo Frame Receipt enable. */
hw_strip_crc : 1, /**< Enable CRC stripping by hardware. */
enable_scatter   : 1, /**< Enable scatter packets rx handler */
-   enable_lro   : 1; /**< Enable LRO */
+   enable_lro   : 1, /**< Enable LRO */
+   enable_sec   : 1; /**< Enable security offload */

Based on the time of integration you may need to update the convert function 
[1] as well.

Indeed.



  };

  /**
@@ -679,8 +682,10 @@ struct rte_eth_txmode {
/**< If set, reject sending out tagged pkts */
hw_vlan_reject_untagged : 1,
/**< If set, reject sending out untagged pkts */
-   hw_vlan_insert_pvid : 1;
+   hw_vlan_insert_pvid : 1,
/**< If set, enable port based VLAN insertion */
+   enable_sec   : 1;
+   /**< Enable security offload */

Considering this enable_sec is an exception in compare to the regular Tx 
offloads which are always enabled and set per queue,
Wouldn't it be better to use the new offloads API [2] for that flag?
I guess so, when the new API will be merged we can revisit the security 
offloads as well.



  };

  /**
@@ -907,6 +912,7 @@ struct rte_eth_conf {  #define
DEV_RX_OFFLOAD_QINQ_STRIP  0x0020  #define
DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x0040
  #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x0080
+#define DEV_RX_OFFLOAD_SECURITY 0x0100

  /**
   * TX offload capabilities of a device.
@@ -926,6 +932,11 @@ struct rte_eth_conf {
  #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO   0x1000/**< Used for
tunneling packet. */
  #define DEV_TX_OFFLOAD_MACSEC_INSERT0x2000
  #define DEV_TX_OFFLOAD_MT_LOCKFREE  0x4000
+#define DEV_TX_OFFLOAD_SECURITY 0x8000

The above flag I understand.


+#define DEV_TX_OFFLOAD_SEC_NEED_MDATA   0x0001
+#define DEV_TX_OFFLOAD_IPSEC_CRYPTO_HW_TRAILER 0x0002
+#define DEV_TX_OFFLOAD_IPSEC_CRYPTO_TSO0x0004
+#define DEV_TX_OFFLOAD_IPSEC_CRYPTO_CKSUM  0x0008

The above 4 flags I don't. doc/comments are missing to describe what exactly 
each feature means.
Also considering those caps may be protocol depended (e.g. PMD can do TSO for 
ipsec but cannot for macsec) isn't it better for those caps to be advertised as 
part of rte_security_capabilities ?
The PMD will report for ethdev layer it support the security offloads, then per 
protocol the above caps will be advertised.

Will be updated, thanks!




  /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the
same
   * tx queue without SW lock.
   */
@@ -1651,6 +1662,9 @@ struct rte_eth_dev {
enum rte_eth_dev_state state; /**< Flag indicating the port state */
} __rte_cache_aligned;

+uint16_t
+rte_eth_dev_get_sec_id(uint8_t port_id);
+
  struct rte_eth_dev_sriov {
uint8_t active;   /**< SRIOV is active with 16, 32 or 64 
pools */
uint8_t nb_q_per_pool;/**< rx queue number per pool */
@@ -1711,6 +1725,8 @@ struct rte_eth_dev_data {
int numa_node;  /**< NUMA node connection */
struct rte_vlan_filter_conf vlan_filter_conf;
/**< VLAN filter configuration. */
+   uint16_t sec_id;
+   /**< security instance identifier */
  };

  /** Device supports hotpl

Re: [dpdk-dev] [PATCH 06/11] ethdev: extend ethdev to support security APIs

[Radu Nicolau]

Hi,


On 9/18/2017 8:57 AM, Jerin Jacob wrote:

-Original Message-

Date: Thu, 14 Sep 2017 13:56:46 +0530
From: Akhil Goyal 
To: dev@dpdk.org
CC: declan.dohe...@intel.com, pablo.de.lara.gua...@intel.com,
  hemant.agra...@nxp.com, radu.nico...@intel.com, bor...@mellanox.com,
  avia...@mellanox.com, tho...@monjalon.net, sandeep.ma...@nxp.com,
  jerin.ja...@caviumnetworks.com
Subject: [PATCH 06/11] ethdev: extend ethdev to support security APIs
X-Mailer: git-send-email 2.9.3

From: Declan Doherty 
  /**
@@ -907,6 +912,7 @@ struct rte_eth_conf {
  #define DEV_RX_OFFLOAD_QINQ_STRIP  0x0020
  #define DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM 0x0040
  #define DEV_RX_OFFLOAD_MACSEC_STRIP 0x0080
+#define DEV_RX_OFFLOAD_SECURITY 0x0100
  
  /**

   * TX offload capabilities of a device.
@@ -926,6 +932,11 @@ struct rte_eth_conf {
  #define DEV_TX_OFFLOAD_GENEVE_TNL_TSO   0x1000/**< Used for tunneling 
packet. */
  #define DEV_TX_OFFLOAD_MACSEC_INSERT0x2000
  #define DEV_TX_OFFLOAD_MT_LOCKFREE  0x4000
+#define DEV_TX_OFFLOAD_SECURITY 0x8000

For the documentation side, Sharing the same views as Shahaf's


+#define DEV_TX_OFFLOAD_SEC_NEED_MDATA   0x0001

If i understand correctly, it is more of a fixup if driver needs.I guess
we can change to negative logic reflect it as offload.
I can make a point for positive logic: if there is no need to update the 
mdata then neither PMD nor app need to do anything. If there is, the PMD 
needs to set the flag and the app needs to check it.





+#define DEV_TX_OFFLOAD_IPSEC_CRYPTO_HW_TRAILER 0x0002
+#define DEV_TX_OFFLOAD_IPSEC_CRYPTO_TSO0x0004
+#define DEV_TX_OFFLOAD_IPSEC_CRYPTO_CKSUM  0x0008
  /**< Multiple threads can invoke rte_eth_tx_burst() concurrently on the same
   * tx queue without SW lock.
   */

The above text should come after DEV_TX_OFFLOAD_MT_LOCKFREE(Position got
changed in this patch)

Thanks, will fix.



@@ -1651,6 +1662,9 @@ struct rte_eth_dev {
enum rte_eth_dev_state state; /**< Flag indicating the port state */
  } __rte_cache_aligned;
  

Re: [dpdk-dev] [PATCH 01/11] lib/rte_security: add security library

[Radu Nicolau]

Hi,

I will address most of the issues in the v2, except the one related to 
the multiprocess issue - we may need more discussions on that.


Thanks for reviewing,

Radu


On 9/18/2017 2:13 PM, Jerin Jacob wrote:

-Original Message-

Date: Thu, 14 Sep 2017 13:56:41 +0530
From: Akhil Goyal 
To: dev@dpdk.org
CC: declan.dohe...@intel.com, pablo.de.lara.gua...@intel.com,
  hemant.agra...@nxp.com, radu.nico...@intel.com, bor...@mellanox.com,
  avia...@mellanox.com, tho...@monjalon.net, sandeep.ma...@nxp.com,
  jerin.ja...@caviumnetworks.com
Subject: [PATCH 01/11] lib/rte_security: add security library
X-Mailer: git-send-email 2.9.3

rte_security library provides APIs for security session
create/free for protocol offload or offloaded crypto
operation to ethernet device.

Overall the API semantic looks good. A few comments inlined.
I think, This patch should split as minimum two. One just the
specification header file and other one implementation.



Signed-off-by: Akhil Goyal 
Signed-off-by: Boris Pismenny 
Signed-off-by: Radu Nicolau 
Signed-off-by: Declan Doherty 
---
+
+#include 
+#include 
+
+#include "rte_security.h"
+#include "rte_security_driver.h"
+
+#define RTE_SECURITY_INSTANCES_BLOCK_ALLOC_SZ  (8)
+
+struct rte_security_ctx {
+   uint16_t id;
+   enum {
+   RTE_SECURITY_INSTANCE_INVALID = 0,

explicit zero is not required.


+   RTE_SECURITY_INSTANCE_VALID
+   } state;
+   void *device;
+   struct rte_security_ops *ops;
+};
+
+
+int
+rte_security_register(uint16_t *id, void *device,
+ struct rte_security_ops *ops)
+{
+   if (max_nb_security_instances == 0) {
+   security_instances = rte_malloc(
+   "rte_security_instances_ops",
+   sizeof(*security_instances) *
+   RTE_SECURITY_INSTANCES_BLOCK_ALLOC_SZ, 0);
+
+   if (security_instances == NULL)
+   return -ENOMEM;
+   max_nb_security_instances =
+   RTE_SECURITY_INSTANCES_BLOCK_ALLOC_SZ;
+   } else if (nb_security_instances >= max_nb_security_instances) {
+   uint16_t *instances = rte_realloc(security_instances,
+   sizeof(struct rte_security_ops *) *
+   (max_nb_security_instances +
+   RTE_SECURITY_INSTANCES_BLOCK_ALLOC_SZ), 0);
+
+   if (instances == NULL)
+   return -ENOMEM;
+
+   max_nb_security_instances +=
+   RTE_SECURITY_INSTANCES_BLOCK_ALLOC_SZ;
+   }
+
+   *id = nb_security_instances++;
+
+   security_instances[*id].id = *id;
+   security_instances[*id].state = RTE_SECURITY_INSTANCE_VALID;
+   security_instances[*id].device = device;
+   security_instances[*id].ops = ops;

This whole thing will break in multi process case where ops needs to
cloned for each process. Check the mempool library as reference.



+
+   return 0;
+}
+
+int
+rte_security_unregister(__rte_unused uint16_t *id)
+{
+   /* To be implemented */

This should implemented before it reaches to master.


+   return 0;
+}
+
+struct rte_security_session *
+int
+rte_security_set_pkt_metadata(uint16_t id,
+ struct rte_security_session *sess,
+ struct rte_mbuf *m, void *params)
+{
+   struct rte_security_ctx *instance;
+
+   RTE_SEC_VALID_ID_OR_ERR_RET(id, -ENODEV);
+   instance = &security_instances[id];
+
+   RTE_FUNC_PTR_OR_ERR_RET(*instance->ops->set_pkt_metadata, -ENOTSUP);

Do you need all this checking for a fastpath function?


+   return instance->ops->set_pkt_metadata(instance->device,
+  sess, m, params);
+}
+
+
+/**
+ * @file rte_security.h
+ *
+ * RTE Security Common Definitions
+ *
+ */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include 
+#include 

Nice to have it in alphabetical order.


+
+/** IPSec protocol mode */
+enum rte_security_ipsec_sa_mode {
+   RTE_SECURITY_IPSEC_SA_MODE_TRANSPORT,
+   /**< IPSec Transport mode */
+   RTE_SECURITY_IPSEC_SA_MODE_TUNNEL,
+   /**< IPSec Tunnel mode */
+};
+
+/** IPSec Protocol */
+enum rte_security_ipsec_sa_protocol {
+   RTE_SECURITY_IPSEC_SA_PROTO_AH,
+   /**< AH protocol */
+   RTE_SECURITY_IPSEC_SA_PROTO_ESP,
+   /**< ESP protocol */
+};
+
+/** IPSEC tunnel type */
+enum rte_security_ipsec_tunnel_type {
+   RTE_SECURITY_IPSEC_TUNNEL_IPV4 = 0,

Explicit zero may not be required.


+   /**< Outer header is IPv4 */
+   RTE_SECURITY_IPSEC_TUNNEL_IPV6,
+   /**< Outer header is IPv6 */
+};



+struct rte_security_ipsec_tunnel_param {
+   enum rte_security_ipsec_tunnel_type type;
+   /**< Tunnel type: IPv4 or IPv6 */
+

Anonymous union, Y

Re: [dpdk-dev] [PATCH v4 4/4] eventdev: Add tests for event eth Rx adapter APIs

[Jerin Jacob]

-Original Message-
> Date: Fri, 22 Sep 2017 02:47:14 +0530
> From: Nikhil Rao 
> To: jerin.ja...@caviumnetworks.com, bruce.richard...@intel.com
> CC: gage.e...@intel.com, dev@dpdk.org, tho...@monjalon.net,
>  harry.van.haa...@intel.com, hemant.agra...@nxp.com, nipun.gu...@nxp.com,
>  narender.vang...@intel.com, erik.g.carri...@intel.com,
>  abhinandan.guj...@intel.com, santosh.shu...@caviumnetworks.com
> Subject: [PATCH v4 4/4] eventdev: Add tests for event eth Rx adapter APIs
> X-Mailer: git-send-email 2.7.4
> 
> Add unit tests for rte_event_eth_rx_adapter_xxx() APIs
> 
> Signed-off-by: Nikhil Rao 
> ---
>  test/test/test_event_eth_rx_adapter.c | 399 
> ++
>  test/test/Makefile|   1 +
>  2 files changed, 400 insertions(+)
>  create mode 100644 test/test/test_event_eth_rx_adapter.c
> 
> diff --git a/test/test/test_event_eth_rx_adapter.c 
> b/test/test/test_event_eth_rx_adapter.c
> new file mode 100644
> index 0..5d448dc27
> --- /dev/null
> +++ b/test/test/test_event_eth_rx_adapter.c
> @@ -0,0 +1,399 @@
> +/*-
> + *   BSD LICENSE
> + *
> + *   Copyright(c) 2017 Intel Corporation. All rights reserved.
> + *
> + *   Redistribution and use in source and binary forms, with or without
> + *   modification, are permitted provided that the following conditions
> + *   are met:
> + *
> + * * Redistributions of source code must retain the above copyright
> + *   notice, this list of conditions and the following disclaimer.
> + * * Redistributions in binary form must reproduce the above copyright
> + *   notice, this list of conditions and the following disclaimer in
> + *   the documentation and/or other materials provided with the
> + *   distribution.
> + * * Neither the name of Intel Corporation nor the names of its
> + *   contributors may be used to endorse or promote products derived
> + *   from this software without specific prior written permission.
> + *
> + *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> + *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> + *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> + *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> + *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> + *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
> + *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
> + *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
> + *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
> + *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
> + *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
> + */
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +
> +#include 
> +
> +#include "test.h"
> +
> +/* i40e limits max to 64 */

This comment could be removed.

> +#define MAX_NUM_RX_QUEUE 64
> +#define NB_MBUFS (8192 * num_ports * MAX_NUM_RX_QUEUE)
> +#define MBUF_CACHE_SIZE  512
> +#define MBUF_PRIV_SIZE   0
> +
> +struct event_eth_rx_adapter_test_params {
> + struct rte_mempool *mp;
> + uint16_t rx_rings, tx_rings;
> + uint32_t caps;
> +};
> +
> +static struct event_eth_rx_adapter_test_params default_params;
> +
> +static int
> +testsuite_setup(void)
> +{
> + int err;
> + err = init_ports(rte_eth_dev_count());
> + TEST_ASSERT(err == 0, "Port initialization failed err %d\n", err);

I guess, We check rte_event_dev_count() >= 1 before proceeding.

> +
> + struct rte_event_dev_config config = {
> + .nb_event_queues = 1,
> + .nb_event_ports = 1,
> + .nb_events_limit  = 4096,
> + .nb_event_queue_flows = 1024,
> + .nb_event_port_dequeue_depth = 16,
> + .nb_event_port_enqueue_depth = 16
> + };
> +
> + err = rte_event_dev_configure(0, &config);
> + TEST_ASSERT(err == 0, "Event device initialization failed err %d\n",
> + err);
> +
> + err = rte_event_eth_rx_adapter_caps_get(0, 0, &default_params.caps);
> + TEST_ASSERT(err == 0, "Failed to get adapter cap err %d\n",
> + err);
> +
> + return err;
> +}
> +

[dpdk-dev] [PATCH v2] hash: fix incorrect eviction counter

[Pablo de Lara]

When adding a new entry in a hash table, there is
a maximum number of evictions that can be
performed. When the counter of these evictions reaches
this maximum, the entry cannot be added, as it is considered
that the algorithm has encountered an infinite loop.

The problem with the current implementation, is that this
counter was declared as a static variable.
If there are multiple threads adding entries in the same table
or in different tables, they should access different counters,
one per core and per table.

Therefore, the variable has been modified to be non-static.

Fixes: 243e93a5046f ("hash: fix unlimited cuckoo path")
Cc: sta...@dpdk.org

Signed-off-by: Pablo de Lara 
---

Changes in v2:
- Used a non-static variable to store counter, instead of
  array in hash structure

 lib/librte_hash/rte_cuckoo_hash.c | 13 ++---
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/lib/librte_hash/rte_cuckoo_hash.c 
b/lib/librte_hash/rte_cuckoo_hash.c
index 87b25c0..e69b911 100644
--- a/lib/librte_hash/rte_cuckoo_hash.c
+++ b/lib/librte_hash/rte_cuckoo_hash.c
@@ -417,9 +417,9 @@ rte_hash_reset(struct rte_hash *h)
 
 /* Search for an entry that can be pushed to its alternative location */
 static inline int
-make_space_bucket(const struct rte_hash *h, struct rte_hash_bucket *bkt)
+make_space_bucket(const struct rte_hash *h, struct rte_hash_bucket *bkt,
+   unsigned int *nr_pushes)
 {
-   static unsigned int nr_pushes;
unsigned i, j;
int ret;
uint32_t next_bucket_idx;
@@ -456,15 +456,14 @@ make_space_bucket(const struct rte_hash *h, struct 
rte_hash_bucket *bkt)
break;
 
/* All entries have been pushed, so entry cannot be added */
-   if (i == RTE_HASH_BUCKET_ENTRIES || nr_pushes > RTE_HASH_MAX_PUSHES)
+   if (i == RTE_HASH_BUCKET_ENTRIES || ++(*nr_pushes) > 
RTE_HASH_MAX_PUSHES)
return -ENOSPC;
 
/* Set flag to indicate that this entry is going to be pushed */
bkt->flag[i] = 1;
 
-   nr_pushes++;
/* Need room in alternative bucket to insert the pushed entry */
-   ret = make_space_bucket(h, next_bkt[i]);
+   ret = make_space_bucket(h, next_bkt[i], nr_pushes);
/*
 * After recursive function.
 * Clear flags and insert the pushed entry
@@ -472,7 +471,6 @@ make_space_bucket(const struct rte_hash *h, struct 
rte_hash_bucket *bkt)
 * or return error
 */
bkt->flag[i] = 0;
-   nr_pushes = 0;
if (ret >= 0) {
next_bkt[i]->sig_alt[ret] = bkt->sig_current[i];
next_bkt[i]->sig_current[ret] = bkt->sig_alt[i];
@@ -515,6 +513,7 @@ __rte_hash_add_key_with_hash(const struct rte_hash *h, 
const void *key,
unsigned n_slots;
unsigned lcore_id;
struct lcore_cache *cached_free_slots = NULL;
+   unsigned int nr_pushes = 0;
 
if (h->add_key == ADD_KEY_MULTIWRITER)
rte_spinlock_lock(h->multiwriter_lock);
@@ -648,7 +647,7 @@ __rte_hash_add_key_with_hash(const struct rte_hash *h, 
const void *key,
 * if successful or return error and
 * store the new slot back in the ring
 */
-   ret = make_space_bucket(h, prim_bkt);
+   ret = make_space_bucket(h, prim_bkt, &nr_pushes);
if (ret >= 0) {
prim_bkt->sig_current[ret] = sig;
prim_bkt->sig_alt[ret] = alt_hash;
-- 
2.9.4

[dpdk-dev] [PATCH] vhost: fix unchecked return value

[Kuba Kozak]

Add return value check for poll() call.

Coverity issue: 140740
Fixes: 59317cef249c ("vhost: allow many vhost-user ports")
Cc: jan.wick...@ericsson.com
Cc: sta...@dpdk.org

Signed-off-by: Kuba Kozak 
---
 lib/librte_vhost/fd_man.c | 5 -
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/lib/librte_vhost/fd_man.c b/lib/librte_vhost/fd_man.c
index 2ceacc9..4c6fed4 100644
--- a/lib/librte_vhost/fd_man.c
+++ b/lib/librte_vhost/fd_man.c
@@ -222,6 +222,7 @@ fdset_event_dispatch(void *arg)
int remove1, remove2;
int need_shrink;
struct fdset *pfdset = arg;
+   int val;
 
if (pfdset == NULL)
return NULL;
@@ -239,7 +240,9 @@ fdset_event_dispatch(void *arg)
numfds = pfdset->num;
pthread_mutex_unlock(&pfdset->fd_mutex);
 
-   poll(pfdset->rwfds, numfds, 1000 /* millisecs */);
+   val = poll(pfdset->rwfds, numfds, 1000 /* millisecs */);
+   if (val < 0)
+   continue;
 
need_shrink = 0;
for (i = 0; i < numfds; i++) {
-- 
2.7.4

[dpdk-dev] [PATCH] net/i40e: fix assignment of enum values

[Michal Jastrzebski]

From: Tomasz Kulasek 

mixed_enums: Mixing enum types enum i40e_vsi_type and enum
 virtchnl_vsi_type for type

Coverity issue 158651
Fixes: a58860f68929 ("net/i40e/base: use new virtchnl header file")
Cc: jingjing...@intel.com
Cc: sta...@dpdk.org

Signed-off-by: Tomasz Kulasek 
---
 drivers/net/i40e/i40e_ethdev_vf.c | 10 +-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/net/i40e/i40e_ethdev_vf.c 
b/drivers/net/i40e/i40e_ethdev_vf.c
index 73c315a..a8d2740 100644
--- a/drivers/net/i40e/i40e_ethdev_vf.c
+++ b/drivers/net/i40e/i40e_ethdev_vf.c
@@ -1295,7 +1295,15 @@ static int i40evf_dev_xstats_get(struct rte_eth_dev *dev,
if (hw->mac.type == I40E_MAC_X722_VF)
vf->flags = I40E_FLAG_RSS_AQ_CAPABLE;
vf->vsi.vsi_id = vf->vsi_res->vsi_id;
-   vf->vsi.type = (enum i40e_vsi_type)vf->vsi_res->vsi_type;
+
+   switch (vf->vsi_res->vsi_type) {
+   case VIRTCHNL_VSI_SRIOV:
+   vf->vsi.type = I40E_VSI_SRIOV;
+   break;
+   default:
+   vf->vsi.type = I40E_VSI_TYPE_UNKNOWN;
+   break;
+   }
vf->vsi.nb_qps = vf->vsi_res->num_queue_pairs;
vf->vsi.adapter = I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 
-- 
1.9.1

[dpdk-dev] [PATCH] net/vmxnet3: fix dereference before null check

[Michal Jastrzebski]

From: Tomasz Kulasek 

Coverity error:

check_after_deref: Null-checking rq suggests that it may be null, but it
   has already been dereferenced on all paths leading to
   the check.

This patch moves NULL checking of "rq" at the very beginning of the path
before any dereference.

Coverity issue: 143468
Fixes: 5aecdc17a97d ("vmxnet3: fix stop/restart")
Cc: yongw...@vmware.com
Cc: sta...@dpdk.org

Signed-off-by: Tomasz Kulasek 
---
 drivers/net/vmxnet3/vmxnet3_rxtx.c | 17 -
 1 file changed, 8 insertions(+), 9 deletions(-)

diff --git a/drivers/net/vmxnet3/vmxnet3_rxtx.c 
b/drivers/net/vmxnet3/vmxnet3_rxtx.c
index d9cf437..4fcceb4 100644
--- a/drivers/net/vmxnet3/vmxnet3_rxtx.c
+++ b/drivers/net/vmxnet3/vmxnet3_rxtx.c
@@ -259,17 +259,16 @@
 {
int i;
vmxnet3_rx_queue_t *rq = rxq;
-   struct vmxnet3_hw *hw = rq->hw;
struct vmxnet3_cmd_ring *ring0, *ring1;
struct vmxnet3_comp_ring *comp_ring;
-   struct vmxnet3_rx_data_ring *data_ring = &rq->data_ring;
int size;
 
-   if (rq != NULL) {
-   /* Release both the cmd_rings mbufs */
-   for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
-   vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
-   }
+   if (rq == NULL)
+   return;
+
+   /* Release both the cmd_rings mbufs */
+   for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
+   vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
 
ring0 = &rq->cmd_ring[0];
ring1 = &rq->cmd_ring[1];
@@ -287,8 +286,8 @@
 
size = sizeof(struct Vmxnet3_RxDesc) * (ring0->size + ring1->size);
size += sizeof(struct Vmxnet3_RxCompDesc) * comp_ring->size;
-   if (VMXNET3_VERSION_GE_3(hw) && rq->data_desc_size)
-   size += rq->data_desc_size * data_ring->size;
+   if (VMXNET3_VERSION_GE_3(rq->hw) && rq->data_desc_size)
+   size += rq->data_desc_size * rq->data_ring.size;
 
memset(ring0->base, 0, size);
 }
-- 
1.9.1

Re: [dpdk-dev] [PATCH v1 01/10] net/i40e: add API to convert VF Id to PF Id

[Hunt, David]

Hi Thomas,


On 22/9/2017 10:56 AM, Thomas Monjalon wrote:

25/08/2017 18:02, David Hunt:

+static inline uint64_t
+vfid_to_pfid_direct(uint8_t port_id, uint64_t vfid)
+{
+   struct rte_eth_dev *dev = &rte_eth_devices[port_id];
+   uint64_t pfid  = (*dev->dev_ops->vfid_to_pfid)(dev, vfid);
+   return pfid;
+}

I would like to comment this API but there is no associated doxygen.


Sure, we'll add Doxygen comments.



If the application is aware of the VFs, it probably already knows
how PF and VF are associated.

Until now, the functions to control VF from PF are driver-specifics.


Working out the relationship between the PF and the VF has turned out to 
be quite a challenge. :)


The application on the guest is aware of the VFs. The application on the 
host is aware of the PF and can access the VFs through the PF. However, 
the application on the host is not aware of how each VF on VM associates 
as which VF on the PF. I.E. the PF needs to know which index in its 
array of VFs the VF in use by the guest application is stored at. This 
is what this additional function is used for. It gives the PF the index 
of the VF in question in its array of VFs. We have researched 
alternative ways to determine this association but this is the only 
method that provides this functionality. Without this the PF does not 
know how each VF is associated with PF.


We also realize that the mac addresses need to be in sync between the 
host and the guest for correct operation of this scheme.


As mentioned in my previous mail, we are working on an updated patch 
set, targeting early next week.


Regards,
Dave.

Re: [dpdk-dev] [PATCH v4 00/41] Introduce NXP DPAA Bus, Mempool and PMD

[Shreyansh Jain]

On Friday 22 September 2017 03:40 AM, Thomas Monjalon wrote:

09/09/2017 13:20, Shreyansh Jain:

DPAA, or Datapath Acceleration Architecture [R2], is a set of hardware
components designed for high-speed network packet processing. This
architecture provides the infrastructure to support simplified sharing of
networking interfaces and accelerators by multiple CPU cores, and the
accelerators themselves.

This patchset introduces the following:
1. DPAA Bus (drivers/bus/dpaa)
  The core of DPAA bus is implemented using 3 main hardware blocks: QMan,
  or Queue Manager; BMan, or Buffer Manager and FMan, or Frame Manager.
  The patches introduce necessary layers to expose the DPAA hardware
  blocks for interfacing with RTE framework.


I guess these are the same blocks as for DPAA2?
They are in drivers/bus/fslmc/
Why introducing yet another bus driver?
The fslmc one was supposed to cover any Freescale (NXP (Qualcomm)) SoC.


Forgot to reply to this in previous email:

No, fslmc is not compatible with DPAA. They are completely different 
architectures.
I am not sure why you have the notion "fslmc one was supposed to cover 
any Freescale (NXP (Qualcomm)) SoC". That is not correct - FSLMC was 
always for supporting DPAA2 which is based on VFIO. DPAA is more closer 
to a platform layout.


And I don't think we should have single "bus/fslmc" just so that it can 
encompass all NXP SoC. I am assuming you didn't mean this :P.





2. DPAA Mempool (drivers/mempool/dpaa)
  BMan, or Buffer Manager, block of DPAA features a hardware offloaded
  mempool. These patches add support for a driver to manage the BMan
  block. This driver allows for mempool creation, deletion, buffer
  acquire and release, as per the RTE APIs.

3. DPAA PMD (drivers/net/dpaa)
  The Poll Mode Driver for DPAA NIC Interfaces.

Patch Layout


01: Add DPAA SoC build configuration
02~16: Add DPAA Bus support and features, incrementally
17: Add Documentation
18~21: Add DPAA Mempool support
22~41: Add PMD and its various features, incrementally


It is a very long series introducing 3 different subsystems.
I think everybody was scared about reviewing it.
Why you did not split it?

Re: [dpdk-dev] [PATCH v6 7/8] mempool: introduce block size align flag

[Hemant Agrawal]

Tested-by:  Hemant Agrawal 

On 9/7/2017 9:00 PM, Santosh Shukla wrote:

Some mempool hw like octeontx/fpa block, demands block size
(/total_elem_sz) aligned object start address.

Introducing an MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS flag.
If this flag is set:
- Align object start address(vaddr) to a multiple of total_elt_sz.
- Allocate one additional object. Additional object is needed to make
  sure that requested 'n' object gets correctly populated.

Example:
- Let's say that we get 'x' size of memory chunk from memzone.
- And application has requested 'n' object from mempool.
- Ideally, we start using objects at start address 0 to...(x-block_sz)
  for n obj.
- Not necessarily first object address i.e. 0 is aligned to block_sz.
- So we derive 'offset' value for block_sz alignment purpose i.e..'off'.
- That 'off' makes sure that start address of object is blk_sz aligned.
- Calculating 'off' may end up sacrificing first block_sz area of
  memzone area x. So total number of the object which can fit in the
  pool area is n-1, Which is incorrect behavior.

Therefore we request one additional object (/block_sz area) from memzone
when MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS flag is set.

Signed-off-by: Santosh Shukla 
Signed-off-by: Jerin Jacob 
---
v5 --> v6:
- Renamed from MEMPOOL_F_BLK_ALIGNED_OBJECTS to
  MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS. (Suggested by Olivier)
- Updated Capability flag descrioption (Suggested by Olivier)

History refer [1]
[1] http://dpdk.org/dev/patchwork/patch/28418/

v4 --> v5:
- Added vaddr in git description of patch (suggested by Olivier)
- Renamed to aligned flag to MEMPOOL_F_BLK_ALIGNED_OBJECTS (suggested by
  Olivier)
Refer [2].
[2] http://dpdk.org/dev/patchwork/patch/27600/

 lib/librte_mempool/rte_mempool.c | 19 ---
 lib/librte_mempool/rte_mempool.h | 12 
 2 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
index 146e38675..decdda3a6 100644
--- a/lib/librte_mempool/rte_mempool.c
+++ b/lib/librte_mempool/rte_mempool.c
@@ -239,10 +239,15 @@ rte_mempool_calc_obj_size(uint32_t elt_size, uint32_t 
flags,
  */
 size_t
 rte_mempool_xmem_size(uint32_t elt_num, size_t total_elt_sz, uint32_t pg_shift,
- __rte_unused unsigned int flags)
+ unsigned int flags)
 {
size_t obj_per_page, pg_num, pg_sz;

+   if (flags & (MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS |
+   MEMPOOL_F_CAPA_PHYS_CONTIG))
+   /* alignment need one additional object */
+   elt_num += 1;
+
if (total_elt_sz == 0)
return 0;

@@ -265,13 +270,18 @@ rte_mempool_xmem_size(uint32_t elt_num, size_t 
total_elt_sz, uint32_t pg_shift,
 ssize_t
 rte_mempool_xmem_usage(__rte_unused void *vaddr, uint32_t elt_num,
size_t total_elt_sz, const phys_addr_t paddr[], uint32_t pg_num,
-   uint32_t pg_shift, __rte_unused unsigned int flags)
+   uint32_t pg_shift, unsigned int flags)
 {
uint32_t elt_cnt = 0;
phys_addr_t start, end;
uint32_t paddr_idx;
size_t pg_sz = (size_t)1 << pg_shift;

+   if (flags & (MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS |
+   MEMPOOL_F_CAPA_PHYS_CONTIG))
+   /* alignment need one additional object */
+   elt_num += 1;
+
/* if paddr is NULL, assume contiguous memory */
if (paddr == NULL) {
start = 0;
@@ -390,7 +400,10 @@ rte_mempool_populate_phys(struct rte_mempool *mp, char 
*vaddr,
memhdr->free_cb = free_cb;
memhdr->opaque = opaque;

-   if (mp->flags & MEMPOOL_F_NO_CACHE_ALIGN)
+   if (mp->flags & MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS)
+   /* align object start address to a multiple of total_elt_sz */
+   off = total_elt_sz - ((uintptr_t)vaddr % total_elt_sz);
+   else if (mp->flags & MEMPOOL_F_NO_CACHE_ALIGN)
off = RTE_PTR_ALIGN_CEIL(vaddr, 8) - vaddr;
else
off = RTE_PTR_ALIGN_CEIL(vaddr, RTE_CACHE_LINE_SIZE) - vaddr;
diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
index 734392556..24195dda0 100644
--- a/lib/librte_mempool/rte_mempool.h
+++ b/lib/librte_mempool/rte_mempool.h
@@ -271,6 +271,18 @@ struct rte_mempool {
  * Note: This flag should not be passed by application.
  */
 #define MEMPOOL_F_CAPA_PHYS_CONTIG 0x0040
+/**
+ * This capability flag is advertised by a mempool handler. Used for a case
+ * where mempool driver wants object start address(vaddr) aligned to block
+ * size(/ total element size).
+ *
+ * Note:
+ * - This flag should not be passed by application.
+ *   Flag used for mempool driver only.
+ * - Mempool driver must also set MEMPOOL_F_CAPA_PHYS_CONTIG flag along with
+ *   MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS.
+ */
+#define MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS 0x0080

 /**
  * @internal When debug is enabled, store some statistics.

Re: [dpdk-dev] [PATCH] net/enic: fix multi-process operation

[John Daley (johndale)]

Hi Ferruh,
-johnd

> -Original Message-
> From: Ferruh Yigit [mailto:ferruh.yi...@intel.com]
> Sent: Tuesday, September 19, 2017 5:59 AM
> To: John Daley (johndale) ; Thomas Monjalon
> 
> Cc: dev@dpdk.org; Sergio Gonzalez Monroy
> ; Adrien Mazarguil
> ; Nelio Laranjeiro
> ; Ananyev, Konstantin
> 
> Subject: Re: [dpdk-dev] [PATCH] net/enic: fix multi-process operation
> 
> On 9/19/2017 6:31 AM, John Daley (johndale) wrote:
> >
> >> -Original Message-
> >> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> >> Sent: Monday, September 18, 2017 3:25 PM
> >> To: John Daley (johndale) 
> >> Cc: Ferruh Yigit ; dev@dpdk.org; Sergio
> >> Gonzalez Monroy 
> >> Subject: Re: [PATCH] net/enic: fix multi-process operation
> >>
> >> 18/09/2017 23:27, Ferruh Yigit:
> >>> On 9/11/2017 7:58 PM, John Daley wrote:
>  - Use rte_malloc() instead of malloc() for the per device 'vdev'
> structure
>    so that it can be shared across processes.
>  - Only initialize the device if the process type is
>  RTE_PROC_PRIMARY
>  - Only allow the primary process to do queue setup, start/stop, promisc
>    allmulticast, mac add/del, mtu.
> >> [...]
>  --- a/drivers/net/enic/enic_ethdev.c
>  +++ b/drivers/net/enic/enic_ethdev.c
>  @@ -142,6 +142,10 @@ enicpmd_dev_filter_ctrl(struct rte_eth_dev
>  *dev,  static void enicpmd_dev_tx_queue_release(void *txq)  {
>   ENICPMD_FUNC_TRACE();
>  +
>  +if (rte_eal_process_type() != RTE_PROC_PRIMARY)
>  +return;
>  +
> >>>
> >>> Hi John,
> >>>
> >>> I am not sure about these updates. Agree that these functions should
> >>> know process type, but all others PMDs don't do this.
> >
> > I looked at mlx5 and it does something similar with its mlx5_is_secondary()
> in functions that modify fields in its shared private structure.
> 
> Right, mlx5 also has these checks.
> 
> >
> >>>
> >>> Added a few more people for comment, but as far I understand its
> >>> application responsibility to NOT call these functions if it is
> >>> secondary process.
> >>>
> >>> For device init/uninit, that is part of eal_init() and have to be
> >>> called both for primary and secondary process and PMD needs to
> >>> protect it, for other functions application's responsibility.
> >
> > I put the checks into the PMD because I didn't want to trust the app and I
> didn't see checks in the library functions. I assumed that is why it was done 
> in
> mlx5. I was afraid that the secondary may try to write fields in the shared
> structure and cause some silent bad behavior, like if secondary sets the MTU
> then the primary does, then secondary assumes it was different than what it
> is, or something like that.
> 
> The set values are in the shared memory, so a variable set by secondary will
> be visible to primary, via versa. Of course a synchronization required
> between primary and secondary processes.
> 
> Also why secondary shouldn't be allowed to do some work, like start/stop
> the port?
> 
> I believe this should be application level concern, at worst libraries but not
> drivers.

I don't disagree, but I would need to verify that relaxing the checks would not 
cause catastrophic errors in case the primary and secondary don't synchronize 
their actions to the PMD. This will take some more testing. The patch I 
provided with the checks is safe as is and so if you could accept it for 17.08 
with the promise that I will work on relaxing the checks in the next release, I 
would appreciate it.
> 
> >>
> >> Yes for now it is the policy.
> >> But it is a gray area and it could be clearer with my "ownership proposal":
> >>http://dpdk.org/ml/archives/dev/2017-September/074656.html
> >> A secondary process could manage the ports it owns.
> >
> > I think the ownership concept would help make DPDK more flexible and
> potentially cleaner. Perhaps ownership checks could be pushed the lib
> functions, like rte_eth_dev_set_mtu(), and remove all such checks in the
> PMD(s).
> >>
> >> Feel free to comment the proposal.

Re: [dpdk-dev] [PATCH v1 0/10] Policy Based Power Control for Guest

[Thomas Monjalon]

22/09/2017 12:28, Hunt, David:
> 
> On 22/9/2017 10:51 AM, Thomas Monjalon wrote:
> > 29/08/2017 15:03, Ananyev, Konstantin:
> >> Hi Dave,
> >>
> >>> This patchset adds the facility for a guest VM to send a policy down to
> >>> the host that will allow the host to scale up/down cpu frequencies
> >>> depending on the policy criteria independently of the DPDK app running in
> >>> the guest.  This differs from the previous vm_power implementation where
> >>> individual scale up/down requests were send from the guest to the host via
> >>> virtio-serial.
> >>>
> >>> It's a modification of the vm_power_manager app that runs in the host, and
> >>> the guest_vm_power_app example app that runs in the guest. This allows the
> >>> guest to send down a policy to the host via virtio-serial, which then 
> >>> allows
> >>> the host to scale up/down based on the criteria in the policy, resulting 
> >>> in
> >>> quicker scale up/down than individual requests coming from the guest.
> >>> It also means that the DPDK application running in the guest does not need
> >>> to be modified in any way, it is unaware that it's cores are being scaled
> >>> up/down, reducing the effort in implementing a power-aware infrastructure.
> >>>
> >>> The usage model is as follows:
> >>> 1. Set up the VF's and assign to the guest in the usual way.
> >>> 2. run vm_power_manager on the host, creating a channel to the guest.
> >>> 3. Start the guest_vm_power_mgr app on the guest, which establishes
> >>> a virtio-serial channel to the host.
> >>> 4. Send down the profile for the guest using the "send_profile now" 
> >>> command.
> >>> There is an example profile hard-coded into guest_vm_power_mgr.
> >>> 5. Stop the guest_vm_power_mgr and run your normal power-unaware 
> >>> application.
> >>> 6. Send traffic into the VFs at varying traffic rates.
> >>> Observe the frequency change on the host (turbostat -i 1)
> >>>
> >>> The sequence of code changes are as follows:
> >>>
> >>> Firstly, two new API calls are added to the ethdev layer
> >>> 1. One to convert a VF id to a PF id. In the patchset
> >>> this id is a MAC address. This is needed so that the host can map the 
> >>> VFs
> >>> in the profile to PF so in can monitor the traffic on the relevant PF 
> >>> at the
> >>> host level.
> >>> 2. The other function is to read the low-level traffic throughput on the 
> >>> NIC.
> >>> Currently this API reads a NIC register for speed, but we are looking 
> >>> at
> >>> using a more generic way to get these stats, suggestions welcome.
> >> Why do you need a server (host) to collect RX/TX statistics for VM?
> >> Such method seems to have a lot of limitations:
> >> - no clear method to identify to which VM that VF belongs.
> >> - rely on HW ability to provide such statistics for PF
> >>(limited HW support).
> >> - wouldn't work if PF is not controlled  by the same DPDK app.
> >> Why not to make  it client(VM) responsibility to collect that statistics 
> >> and
> >> periodically send it to the server?
> >> Then server just will have to process that data and make decision.
> > Any progress Dave?
> >
> > You have another series "turbo boost API". Does it depends on this one?
> 
> Hi Thomas,
> 
> We're still working on updates based on Konstantin's feedback above, and 
> hope to have a new patch set submitted to the mailing list early next 
> week. This will remove the ethdev layer changes, and uses pre-existing 
> stats-api.
> 
> In relation to the Turbo patch, they are still independent, but when we 
> have the next revision of the Policy patch submitted, I'll do a new 
> version of the Turbo patch so that it can be applied on top of the 
> policy patch.

OK, thanks

If the turbo patch is independent, I can push it now?

[dpdk-dev] [PATCH] examples/vhost_scsi: fix buffer not terminated

[Michal Jastrzebski]

From: Jacek Piasecki 

Fix size of buffer in strcpy. There was possible to get
not terminated string after copy operation.

Coverity issue: 158631
Fixes: db75c7af19bb ("examples/vhost_scsi: introduce a new sample app")
Cc: changpeng@intel.com
Cc: sta...@dpdk.org

Signed-off-by: Jacek Piasecki 
---
 examples/vhost_scsi/scsi.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/examples/vhost_scsi/scsi.c b/examples/vhost_scsi/scsi.c
index 54d3104..de9639a 100644
--- a/examples/vhost_scsi/scsi.c
+++ b/examples/vhost_scsi/scsi.c
@@ -307,7 +307,8 @@
strncpy((char *)inqdata->t10_vendor_id, "INTEL", 8);
 
/* PRODUCT IDENTIFICATION */
-   strncpy((char *)inqdata->product_id, bdev->product_name, 16);
+   strncpy((char *)inqdata->product_id, bdev->product_name,
+   ARRAY_SIZE(inqdata->product_id) - 1);
 
/* PRODUCT REVISION LEVEL */
strncpy((char *)inqdata->product_rev, "0001", 4);
-- 
1.9.1

[dpdk-dev] [PATCH] net/vmxnet3: fix dereference before null check

[Michal Jastrzebski]

From: Tomasz Kulasek 

Coverity error:

check_after_deref: Null-checking rq suggests that it may be null, but it
   has already been dereferenced on all paths leading to
   the check.

This patch moves NULL checking of "rq" at the very beginning of the path
before any dereference.

Coverity issue: 143468
Fixes: 5aecdc17a97d ("vmxnet3: fix stop/restart")
Cc: yongw...@vmware.com
Cc: sta...@dpdk.org

Signed-off-by: Tomasz Kulasek 
---
 drivers/net/vmxnet3/vmxnet3_rxtx.c | 17 -
 1 file changed, 8 insertions(+), 9 deletions(-)

diff --git a/drivers/net/vmxnet3/vmxnet3_rxtx.c 
b/drivers/net/vmxnet3/vmxnet3_rxtx.c
index d9cf437..4fcceb4 100644
--- a/drivers/net/vmxnet3/vmxnet3_rxtx.c
+++ b/drivers/net/vmxnet3/vmxnet3_rxtx.c
@@ -259,17 +259,16 @@
 {
int i;
vmxnet3_rx_queue_t *rq = rxq;
-   struct vmxnet3_hw *hw = rq->hw;
struct vmxnet3_cmd_ring *ring0, *ring1;
struct vmxnet3_comp_ring *comp_ring;
-   struct vmxnet3_rx_data_ring *data_ring = &rq->data_ring;
int size;
 
-   if (rq != NULL) {
-   /* Release both the cmd_rings mbufs */
-   for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
-   vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
-   }
+   if (rq == NULL)
+   return;
+
+   /* Release both the cmd_rings mbufs */
+   for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
+   vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
 
ring0 = &rq->cmd_ring[0];
ring1 = &rq->cmd_ring[1];
@@ -287,8 +286,8 @@
 
size = sizeof(struct Vmxnet3_RxDesc) * (ring0->size + ring1->size);
size += sizeof(struct Vmxnet3_RxCompDesc) * comp_ring->size;
-   if (VMXNET3_VERSION_GE_3(hw) && rq->data_desc_size)
-   size += rq->data_desc_size * data_ring->size;
+   if (VMXNET3_VERSION_GE_3(rq->hw) && rq->data_desc_size)
+   size += rq->data_desc_size * rq->data_ring.size;
 
memset(ring0->base, 0, size);
 }
-- 
1.9.1

[dpdk-dev] [PATCH] examples/vhost_scsi: fix buffer not terminated

[Michal Jastrzebski]

From: Jacek Piasecki 

Fix size of buffer in strcpy. There was possible to get
not terminated string after copy operation.

Coverity issue: 158631
Fixes: db75c7af19bb ("examples/vhost_scsi: introduce a new sample app")
Cc: changpeng@intel.com
Cc: sta...@dpdk.org

Signed-off-by: Jacek Piasecki 
---
 examples/vhost_scsi/scsi.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/examples/vhost_scsi/scsi.c b/examples/vhost_scsi/scsi.c
index 54d3104..de9639a 100644
--- a/examples/vhost_scsi/scsi.c
+++ b/examples/vhost_scsi/scsi.c
@@ -307,7 +307,8 @@
strncpy((char *)inqdata->t10_vendor_id, "INTEL", 8);
 
/* PRODUCT IDENTIFICATION */
-   strncpy((char *)inqdata->product_id, bdev->product_name, 16);
+   strncpy((char *)inqdata->product_id, bdev->product_name,
+   ARRAY_SIZE(inqdata->product_id) - 1);
 
/* PRODUCT REVISION LEVEL */
strncpy((char *)inqdata->product_rev, "0001", 4);
-- 
1.9.1

[dpdk-dev] [PATCH] examples/vhost_scsi: fix buffer not terminated

[Michal Jastrzebski]

From: Jacek Piasecki 

Fix size of buffer in strcpy. There was possible to get
not terminated string after copy operation.

Coverity issue: 158629
Fixes: db75c7af19bb ("examples/vhost_scsi: introduce a new sample app")
Cc: changpeng@intel.com
Cc: sta...@dpdk.org

Signed-off-by: Jacek Piasecki 
---
 examples/vhost_scsi/vhost_scsi.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/examples/vhost_scsi/vhost_scsi.c b/examples/vhost_scsi/vhost_scsi.c
index b4f1f8d..b1a8c93 100644
--- a/examples/vhost_scsi/vhost_scsi.c
+++ b/examples/vhost_scsi/vhost_scsi.c
@@ -186,8 +186,9 @@ static uint64_t gpa_to_vva(int vid, uint64_t gpa)
if (!bdev)
return NULL;
 
-   strncpy(bdev->name, bdev_name, sizeof(bdev->name));
-   strncpy(bdev->product_name, bdev_serial, sizeof(bdev->product_name));
+   strncpy(bdev->name, bdev_name, sizeof(bdev->name) - 1);
+   strncpy(bdev->product_name, bdev_serial,
+   sizeof(bdev->product_name) - 1);
bdev->blocklen = blk_size;
bdev->blockcnt = blk_cnt;
bdev->write_cache = wce_enable;
-- 
1.9.1

Re: [dpdk-dev] [PATCH v1 0/10] Policy Based Power Control for Guest

[Hunt, David]

On 22/9/2017 2:03 PM, Thomas Monjalon wrote:

22/09/2017 12:28, Hunt, David:

On 22/9/2017 10:51 AM, Thomas Monjalon wrote:

29/08/2017 15:03, Ananyev, Konstantin:

Hi Dave,


This patchset adds the facility for a guest VM to send a policy down to
the host that will allow the host to scale up/down cpu frequencies
depending on the policy criteria independently of the DPDK app running in
the guest.  This differs from the previous vm_power implementation where
individual scale up/down requests were send from the guest to the host via
virtio-serial.

It's a modification of the vm_power_manager app that runs in the host, and
the guest_vm_power_app example app that runs in the guest. This allows the
guest to send down a policy to the host via virtio-serial, which then allows
the host to scale up/down based on the criteria in the policy, resulting in
quicker scale up/down than individual requests coming from the guest.
It also means that the DPDK application running in the guest does not need
to be modified in any way, it is unaware that it's cores are being scaled
up/down, reducing the effort in implementing a power-aware infrastructure.

The usage model is as follows:
1. Set up the VF's and assign to the guest in the usual way.
2. run vm_power_manager on the host, creating a channel to the guest.
3. Start the guest_vm_power_mgr app on the guest, which establishes
 a virtio-serial channel to the host.
4. Send down the profile for the guest using the "send_profile now" command.
 There is an example profile hard-coded into guest_vm_power_mgr.
5. Stop the guest_vm_power_mgr and run your normal power-unaware application.
6. Send traffic into the VFs at varying traffic rates.
 Observe the frequency change on the host (turbostat -i 1)

The sequence of code changes are as follows:

Firstly, two new API calls are added to the ethdev layer
1. One to convert a VF id to a PF id. In the patchset
 this id is a MAC address. This is needed so that the host can map the VFs
 in the profile to PF so in can monitor the traffic on the relevant PF at 
the
 host level.
2. The other function is to read the low-level traffic throughput on the NIC.
 Currently this API reads a NIC register for speed, but we are looking at
 using a more generic way to get these stats, suggestions welcome.

Why do you need a server (host) to collect RX/TX statistics for VM?
Such method seems to have a lot of limitations:
- no clear method to identify to which VM that VF belongs.
- rely on HW ability to provide such statistics for PF
(limited HW support).
- wouldn't work if PF is not controlled  by the same DPDK app.
Why not to make  it client(VM) responsibility to collect that statistics and
periodically send it to the server?
Then server just will have to process that data and make decision.

Any progress Dave?

You have another series "turbo boost API". Does it depends on this one?

Hi Thomas,

We're still working on updates based on Konstantin's feedback above, and
hope to have a new patch set submitted to the mailing list early next
week. This will remove the ethdev layer changes, and uses pre-existing
stats-api.

In relation to the Turbo patch, they are still independent, but when we
have the next revision of the Policy patch submitted, I'll do a new
version of the Turbo patch so that it can be applied on top of the
policy patch.

OK, thanks

If the turbo patch is independent, I can push it now?


Yes, absolutely. And I can then ensure the next version of the policy 
patch-set applies on top of it.


Regards,
Dave.

Re: [dpdk-dev] [PATCH v4 00/41] Introduce NXP DPAA Bus, Mempool and PMD

[Thomas Monjalon]

22/09/2017 15:06, Shreyansh Jain:
> On Friday 22 September 2017 03:40 AM, Thomas Monjalon wrote:
> > 09/09/2017 13:20, Shreyansh Jain:
> >> DPAA, or Datapath Acceleration Architecture [R2], is a set of hardware
> >> components designed for high-speed network packet processing. This
> >> architecture provides the infrastructure to support simplified sharing of
> >> networking interfaces and accelerators by multiple CPU cores, and the
> >> accelerators themselves.
> >>
> >> This patchset introduces the following:
> >> 1. DPAA Bus (drivers/bus/dpaa)
> >>   The core of DPAA bus is implemented using 3 main hardware blocks: QMan,
> >>   or Queue Manager; BMan, or Buffer Manager and FMan, or Frame Manager.
> >>   The patches introduce necessary layers to expose the DPAA hardware
> >>   blocks for interfacing with RTE framework.
> > 
> > I guess these are the same blocks as for DPAA2?
> > They are in drivers/bus/fslmc/
> > Why introducing yet another bus driver?
> > The fslmc one was supposed to cover any Freescale (NXP (Qualcomm)) SoC.
> 
> Forgot to reply to this in previous email:
> 
> No, fslmc is not compatible with DPAA. They are completely different 
> architectures.
> I am not sure why you have the notion "fslmc one was supposed to cover 
> any Freescale (NXP (Qualcomm)) SoC". That is not correct - FSLMC was 
> always for supporting DPAA2 which is based on VFIO. DPAA is more closer 
> to a platform layout.
> 
> And I don't think we should have single "bus/fslmc" just so that it can 
> encompass all NXP SoC. I am assuming you didn't mean this :P.

At the beginning of fslmc work, I had understood that every NXP SoC were
connecting components with the same principle which we could call the
"Freescale bus".
Then you came with this bus named bus/fslmc, not bus/dpaa2.
Now I am confused. What is the exact scope of fslmc? Is it just DPAA2?

Re: [dpdk-dev] [PATCH] net/vmxnet3: fix dereference before null check

[Jastrzebski, MichalX K]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Michal Jastrzebski
> Sent: Friday, September 22, 2017 3:08 PM
> To: y...@fridaylinux.org; maxime.coque...@redhat.com
> Cc: dev@dpdk.org; Jain, Deepak K ; Kulasek,
> TomaszX ; yongw...@vmware.com;
> sta...@dpdk.org
> Subject: [dpdk-dev] [PATCH] net/vmxnet3: fix dereference before null check
> 
> From: Tomasz Kulasek 
> 
> Coverity error:
> 
> check_after_deref: Null-checking rq suggests that it may be null, but it
>has already been dereferenced on all paths leading to
>the check.
> 
> This patch moves NULL checking of "rq" at the very beginning of the path
> before any dereference.
> 
> Coverity issue: 143468
> Fixes: 5aecdc17a97d ("vmxnet3: fix stop/restart")
> Cc: yongw...@vmware.com
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Tomasz Kulasek 
> ---
>  drivers/net/vmxnet3/vmxnet3_rxtx.c | 17 -
>  1 file changed, 8 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/net/vmxnet3/vmxnet3_rxtx.c
> b/drivers/net/vmxnet3/vmxnet3_rxtx.c
> index d9cf437..4fcceb4 100644
> --- a/drivers/net/vmxnet3/vmxnet3_rxtx.c
> +++ b/drivers/net/vmxnet3/vmxnet3_rxtx.c
> @@ -259,17 +259,16 @@
>  {
>   int i;
>   vmxnet3_rx_queue_t *rq = rxq;
> - struct vmxnet3_hw *hw = rq->hw;
>   struct vmxnet3_cmd_ring *ring0, *ring1;
>   struct vmxnet3_comp_ring *comp_ring;
> - struct vmxnet3_rx_data_ring *data_ring = &rq->data_ring;
>   int size;
> 
> - if (rq != NULL) {
> - /* Release both the cmd_rings mbufs */
> - for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
> - vmxnet3_rx_cmd_ring_release_mbufs(&rq-
> >cmd_ring[i]);
> - }
> + if (rq == NULL)
> + return;
> +
> + /* Release both the cmd_rings mbufs */
> + for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
> + vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
> 
>   ring0 = &rq->cmd_ring[0];
>   ring1 = &rq->cmd_ring[1];
> @@ -287,8 +286,8 @@
> 
>   size = sizeof(struct Vmxnet3_RxDesc) * (ring0->size + ring1->size);
>   size += sizeof(struct Vmxnet3_RxCompDesc) * comp_ring->size;
> - if (VMXNET3_VERSION_GE_3(hw) && rq->data_desc_size)
> - size += rq->data_desc_size * data_ring->size;
> + if (VMXNET3_VERSION_GE_3(rq->hw) && rq->data_desc_size)
> + size += rq->data_desc_size * rq->data_ring.size;
> 
>   memset(ring0->base, 0, size);
>  }
> --
> 1.9.1

I am sorry, please ignore this mail.

Re: [dpdk-dev] [PATCH] examples/vhost_scsi: fix buffer not terminated

[Jastrzebski, MichalX K]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Michal Jastrzebski
> Sent: Friday, September 22, 2017 3:08 PM
> To: y...@fridaylinux.org; maxime.coque...@redhat.com
> Cc: dev@dpdk.org; Jain, Deepak K ; Piasecki,
> JacekX ; Liu, Changpeng
> ; sta...@dpdk.org
> Subject: [dpdk-dev] [PATCH] examples/vhost_scsi: fix buffer not terminated
> 
> From: Jacek Piasecki 
> 
> Fix size of buffer in strcpy. There was possible to get
> not terminated string after copy operation.
> 
> Coverity issue: 158631
> Fixes: db75c7af19bb ("examples/vhost_scsi: introduce a new sample app")
> Cc: changpeng@intel.com
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Jacek Piasecki 
> ---
>  examples/vhost_scsi/scsi.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/examples/vhost_scsi/scsi.c b/examples/vhost_scsi/scsi.c
> index 54d3104..de9639a 100644
> --- a/examples/vhost_scsi/scsi.c
> +++ b/examples/vhost_scsi/scsi.c
> @@ -307,7 +307,8 @@
>   strncpy((char *)inqdata->t10_vendor_id, "INTEL", 8);
> 
>   /* PRODUCT IDENTIFICATION */
> - strncpy((char *)inqdata->product_id, bdev->product_name,
> 16);
> + strncpy((char *)inqdata->product_id, bdev->product_name,
> + ARRAY_SIZE(inqdata->product_id) - 1);
> 
>   /* PRODUCT REVISION LEVEL */
>   strncpy((char *)inqdata->product_rev, "0001", 4);
> --
> 1.9.1

I am sorry, please ignore this mail - sent two times.

Re: [dpdk-dev] [PATCH v3 4/8] ethdev: add GTP items to support flow API

[Adrien Mazarguil]

On Sat, Sep 23, 2017 at 06:35:10AM +0800, Beilei Xing wrote:
> This patch adds GTP, GTPC and GTPU items for
> generic flow API, and also exposes item fields
> through the flow command.
> 
> Signed-off-by: Beilei Xing 

> diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
> +static const enum index item_gtp[] = {
> + ITEM_GTP_TEID,
> + ITEM_NEXT,
> + ZERO,
> +};
> +
> +static const enum index item_gtpc[] = {
> + ITEM_GTP_TEID,
> + ITEM_NEXT,
> + ZERO,
> +};
> +
> +static const enum index item_gtpu[] = {
> + ITEM_GTP_TEID,
> + ITEM_NEXT,
> + ZERO,
> +};


Only item_gtp[] needs to be defined. GTPC and GTPU should use the same
array. Apart from that:

Acked-by: Adrien Mazarguil 

-- 
Adrien Mazarguil
6WIND

Re: [dpdk-dev] [dpdk-stable] [PATCH] devtools: fix version search with git < 2.7.0

[Thomas Monjalon]

15/09/2017 11:24, Nélio Laranjeiro:
> On Tue, Sep 05, 2017 at 12:05:32AM +0200, Thomas Monjalon wrote:
> > The script git-log-fixes.sh (used in check-git-log.sh) looks
> > for git tags to find the version where a bug is introduced.
> > 
> > In DPDK 17.08, the script has been fixed to ignore tags from
> > non current branch.
> > It was using the option --merged which was introduced in git 2.7.0.
> > 
> > As git 2.7.0 is not so old, a fallback is provided for some years.
> > 
> > The fallback is replacing the tag --merged option by a branch filter.
> > If the tag is found in the branch, the branch name is replaced
> > by the tag.
> > 
> > This script could be improved to allow using another reference branch,
> > instead of hard coding HEAD branch (the current one).
> > 
> > Fixes: 26857dabb3c9 ("devtools: ignore non merged tags for backport")
> > Cc: sta...@dpdk.org
> > 
> > Signed-off-by: Thomas Monjalon 
> Tested-by: Nelio Laranjeiro 
> 
> Solves the issue.

Applied

Re: [dpdk-dev] [PATCH 0/3] update git tree maintainers

[Thomas Monjalon]

> Thomas Monjalon (3):
>   maintainers: separate main tree and devtools maintenance
>   maintainers: add Yuanhan for stable branches
>   maintainers: add Ferruh for main branch

Applied

Re: [dpdk-dev] [PATCH v2] hash: fix incorrect eviction counter

[Bruce Richardson]

On Fri, Sep 22, 2017 at 05:25:43AM +0100, Pablo de Lara wrote:
> When adding a new entry in a hash table, there is
> a maximum number of evictions that can be
> performed. When the counter of these evictions reaches
> this maximum, the entry cannot be added, as it is considered
> that the algorithm has encountered an infinite loop.
> 
> The problem with the current implementation, is that this
> counter was declared as a static variable.
> If there are multiple threads adding entries in the same table
> or in different tables, they should access different counters,
> one per core and per table.
> 
> Therefore, the variable has been modified to be non-static.
> 
> Fixes: 243e93a5046f ("hash: fix unlimited cuckoo path")
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Pablo de Lara 
> ---
Acked-by: Bruce Richardson 

Re: [dpdk-dev] [PATCH v4 00/41] Introduce NXP DPAA Bus, Mempool and PMD

[Shreyansh Jain]

Hello Thomas,

> -Original Message-
> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> Sent: Friday, September 22, 2017 6:43 PM
> To: Shreyansh Jain 
> Cc: dev@dpdk.org; ferruh.yi...@intel.com; Hemant Agrawal
> 
> Subject: Re: [PATCH v4 00/41] Introduce NXP DPAA Bus, Mempool and PMD
> 
> 22/09/2017 15:06, Shreyansh Jain:
> > On Friday 22 September 2017 03:40 AM, Thomas Monjalon wrote:
> > > 09/09/2017 13:20, Shreyansh Jain:
> > >> DPAA, or Datapath Acceleration Architecture [R2], is a set of hardware
> > >> components designed for high-speed network packet processing. This
> > >> architecture provides the infrastructure to support simplified sharing
> of
> > >> networking interfaces and accelerators by multiple CPU cores, and the
> > >> accelerators themselves.
> > >>
> > >> This patchset introduces the following:
> > >> 1. DPAA Bus (drivers/bus/dpaa)
> > >>   The core of DPAA bus is implemented using 3 main hardware blocks:
> QMan,
> > >>   or Queue Manager; BMan, or Buffer Manager and FMan, or Frame Manager.
> > >>   The patches introduce necessary layers to expose the DPAA hardware
> > >>   blocks for interfacing with RTE framework.
> > >
> > > I guess these are the same blocks as for DPAA2?
> > > They are in drivers/bus/fslmc/
> > > Why introducing yet another bus driver?
> > > The fslmc one was supposed to cover any Freescale (NXP (Qualcomm)) SoC.
> >
> > Forgot to reply to this in previous email:
> >
> > No, fslmc is not compatible with DPAA. They are completely different
> > architectures.
> > I am not sure why you have the notion "fslmc one was supposed to cover
> > any Freescale (NXP (Qualcomm)) SoC". That is not correct - FSLMC was
> > always for supporting DPAA2 which is based on VFIO. DPAA is more closer
> > to a platform layout.
> >
> > And I don't think we should have single "bus/fslmc" just so that it can
> > encompass all NXP SoC. I am assuming you didn't mean this :P.
> 
> At the beginning of fslmc work, I had understood that every NXP SoC were
> connecting components with the same principle which we could call the
> "Freescale bus".
> Then you came with this bus named bus/fslmc, not bus/dpaa2.
> Now I am confused. What is the exact scope of fslmc? Is it just DPAA2?

My memory is poor. I will have to look through the old emails what happened - 
but I recall there was a discussion in initial phases about the naming. "fslmc" 
came out as a name that is what is the real name of the DPAA2 bus. There was 
initial a confusion if name of bus in Linux Kernel should match or not - but, 
we realized that bus is *not* device and device name is "dpaa2".

As for whether fslmc would cover multiple SoC - that is still true. There are 
multiple SoCs within the DPAA2 umbrella. LS20XX, LS108X series and some more - 
all of which use the FSLMC bus (DPAA2 architecture, on FSLMC bus, having 
'dpaa2' devices).

There is another architecture, an old one, which are still popular. This is 
platform type bus which is aptly named 'dpaa' - and here the confusion of bus 
name and device doesn't appear. (DPAA bus, using DPAA architecture, exposing 
'dpaa' devices).

Exact scope of FSLMC is just DPAA2 architecture based SoCs. There are many here 
with new coming up.
Exact scope of DPAA bus is just DPAA architecture based SoCs. There are many 
here.

Does this clear your doubt to some extent?

Re: [dpdk-dev] [PATCH v4 00/41] Introduce NXP DPAA Bus, Mempool and PMD

[Thomas Monjalon]

22/09/2017 16:00, Shreyansh Jain:
> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > At the beginning of fslmc work, I had understood that every NXP SoC were
> > connecting components with the same principle which we could call the
> > "Freescale bus".
> > Then you came with this bus named bus/fslmc, not bus/dpaa2.
> > Now I am confused. What is the exact scope of fslmc? Is it just DPAA2?
> 
> My memory is poor. I will have to look through the old emails what happened - 
> but I recall there was a discussion in initial phases about the naming. 
> "fslmc" came out as a name that is what is the real name of the DPAA2 bus. 
> There was initial a confusion if name of bus in Linux Kernel should match or 
> not - but, we realized that bus is *not* device and device name is "dpaa2".
> 
> As for whether fslmc would cover multiple SoC - that is still true. There are 
> multiple SoCs within the DPAA2 umbrella. LS20XX, LS108X series and some more 
> - all of which use the FSLMC bus (DPAA2 architecture, on FSLMC bus, having 
> 'dpaa2' devices).
> 
> There is another architecture, an old one, which are still popular. This is 
> platform type bus which is aptly named 'dpaa' - and here the confusion of bus 
> name and device doesn't appear. (DPAA bus, using DPAA architecture, exposing 
> 'dpaa' devices).
> 
> Exact scope of FSLMC is just DPAA2 architecture based SoCs. There are many 
> here with new coming up.
> Exact scope of DPAA bus is just DPAA architecture based SoCs. There are many 
> here.
> 
> Does this clear your doubt to some extent?

Yes it is a lot clearer! Thanks

Now that I better understand, I think flsmc bus should have been named
dpaa2 bus. Is it too late?

Re: [dpdk-dev] [PATCH v2 0/4] add per-core Turbo Boost capability

[Thomas Monjalon]

13/09/2017 12:44, David Hunt:
> Recent generations of the Intel® Xeon® family processors allow Turbo Boost
> to be enabled/disabled on a per-core basis.
> 
> This patch set introduces additional API calls to the librte_power library
> to allow users to enable/disable Turbo Boost on particular cores.

Applied, thanks

[dpdk-dev] [PATCH v2] eal: fix resource leak

[Daniel Mrzyglod]

Memory allocated in strdup is not free.

Coverity issue: 143257
Fixes: d8a2bc71dfc2 ("log: remove app path from syslog id")
Cc: tho...@monjalon.net

Signed-off-by: Daniel Mrzyglod 
---
v2:
* Fix due to compilation errors

 lib/librte_eal/linuxapp/eal/eal.c | 18 +-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/lib/librte_eal/linuxapp/eal/eal.c 
b/lib/librte_eal/linuxapp/eal/eal.c
index 48f12f4..a7df566 100644
--- a/lib/librte_eal/linuxapp/eal/eal.c
+++ b/lib/librte_eal/linuxapp/eal/eal.c
@@ -751,7 +751,7 @@ rte_eal_init(int argc, char **argv)
int i, fctret, ret;
pthread_t thread_id;
static rte_atomic32_t run_once = RTE_ATOMIC32_INIT(0);
-   const char *logid;
+   char *logid;
char cpuset[RTE_CPU_AFFINITY_STR_LEN];
char thread_name[RTE_MAX_THREAD_NAME_LEN];
 
@@ -781,6 +781,7 @@ rte_eal_init(int argc, char **argv)
if (rte_eal_cpu_init() < 0) {
rte_eal_init_alert("Cannot detect lcores.");
rte_errno = ENOTSUP;
+   free(logid);
return -1;
}
 
@@ -789,6 +790,7 @@ rte_eal_init(int argc, char **argv)
rte_eal_init_alert("Invalid 'command line' arguments.");
rte_errno = EINVAL;
rte_atomic32_clear(&run_once);
+   free(logid);
return -1;
}
 
@@ -799,6 +801,7 @@ rte_eal_init(int argc, char **argv)
rte_eal_init_alert("Cannot get hugepage information.");
rte_errno = EACCES;
rte_atomic32_clear(&run_once);
+   free(logid);
return -1;
}
 
@@ -826,6 +829,7 @@ rte_eal_init(int argc, char **argv)
rte_eal_init_alert("Cannot init logging.");
rte_errno = ENOMEM;
rte_atomic32_clear(&run_once);
+   free(logid);
return -1;
}
 
@@ -834,6 +838,7 @@ rte_eal_init(int argc, char **argv)
rte_eal_init_alert("Cannot init VFIO\n");
rte_errno = EAGAIN;
rte_atomic32_clear(&run_once);
+   free(logid);
return -1;
}
 #endif
@@ -841,6 +846,7 @@ rte_eal_init(int argc, char **argv)
if (rte_eal_memory_init() < 0) {
rte_eal_init_alert("Cannot init memory\n");
rte_errno = ENOMEM;
+   free(logid);
return -1;
}
 
@@ -850,24 +856,28 @@ rte_eal_init(int argc, char **argv)
if (rte_eal_memzone_init() < 0) {
rte_eal_init_alert("Cannot init memzone\n");
rte_errno = ENODEV;
+   free(logid);
return -1;
}
 
if (rte_eal_tailqs_init() < 0) {
rte_eal_init_alert("Cannot init tail queues for objects\n");
rte_errno = EFAULT;
+   free(logid);
return -1;
}
 
if (rte_eal_alarm_init() < 0) {
rte_eal_init_alert("Cannot init interrupt-handling thread\n");
/* rte_eal_alarm_init sets rte_errno on failure. */
+   free(logid);
return -1;
}
 
if (rte_eal_timer_init() < 0) {
rte_eal_init_alert("Cannot init HPET or TSC timers\n");
rte_errno = ENOTSUP;
+   free(logid);
return -1;
}
 
@@ -886,17 +896,20 @@ rte_eal_init(int argc, char **argv)
 
if (rte_eal_intr_init() < 0) {
rte_eal_init_alert("Cannot init interrupt-handling thread\n");
+   free(logid);
return -1;
}
 
if (eal_option_device_parse()) {
rte_errno = ENODEV;
+   free(logid);
return -1;
}
 
if (rte_bus_scan()) {
rte_eal_init_alert("Cannot scan the buses for devices\n");
rte_errno = ENODEV;
+   free(logid);
return -1;
}
 
@@ -941,6 +954,7 @@ rte_eal_init(int argc, char **argv)
if (ret) {
rte_eal_init_alert("rte_service_init() failed\n");
rte_errno = ENOEXEC;
+   free(logid);
return -1;
}
 
@@ -948,6 +962,7 @@ rte_eal_init(int argc, char **argv)
if (rte_bus_probe()) {
rte_eal_init_alert("Cannot probe devices\n");
rte_errno = ENOTSUP;
+   free(logid);
return -1;
}
 
@@ -957,6 +972,7 @@ rte_eal_init(int argc, char **argv)
ret = rte_service_start_with_defaults();
if (ret < 0 && ret != -ENOTSUP) {
rte_errno = ENOEXEC;
+   free(logid);
return -1;
}
 
-- 
2.7.4

[dpdk-dev] [PATCH v8] ether: add support for vtune task tracing

[ilia . kurakin]

From: Ilia Kurakin 

The patch simplifies DPDK applications analysis for developers which use
Intel® VTune Amplifier.

The empty cycles are such iterations that yielded no RX packets. As far as
DPDK is running in poll mode, wasting cycles is equal to wasting CPU time.
Tracing such iterations can identify that device is underutilized. Tracing
empty cycles becomes even more critical if a system uses a lot of Ethernet
ports.

The patch gives possibility to analyze empty cycles without changing
application code. All needs to be done is just to reconfigure and rebuild
the DPDK itself with CONFIG_RTE_ETHDEV_PROFILE_ITT_WASTED_RX_ITERATIONS
enbled. The important thing here is that this does not affect DPDK code.
The profiling code is not being compiled if user does not specify config
flag.

The patch provides common way to inject RX queues profiling and VTune
specific implementation.

Signed-off-by: Ilia Kurakin 

---

-V2 change:
ITT tasks collection is moved to rx callback

-V3 change:
rte_ethdev_profile.c created, all profile specific code moved there.

Added generic profile function

-V4 change:
checkpatch issues fixed

Added documentation topic

-V5 change:
Documentation fixes

-V6 change:
Documentation changes are moved to different patch

-V7 change:
Documentation and code changes are merged to a single patch.
Added detailed description.

-V8 change:
Added return code handling for Rx queues profiling init.


 config/common_base|   1 +
 doc/guides/prog_guide/profile_app.rst |  37 +++-
 lib/librte_ether/Makefile |   1 +
 lib/librte_ether/ethdev_profile.c | 166 ++
 lib/librte_ether/ethdev_profile.h |  56 
 lib/librte_ether/rte_ethdev.c |  11 +++
 6 files changed, 271 insertions(+), 1 deletion(-)
 create mode 100644 lib/librte_ether/ethdev_profile.c
 create mode 100644 lib/librte_ether/ethdev_profile.h

diff --git a/config/common_base b/config/common_base
index 5e97a08..12f6be9 100644
--- a/config/common_base
+++ b/config/common_base
@@ -136,6 +136,7 @@ CONFIG_RTE_MAX_QUEUES_PER_PORT=1024
 CONFIG_RTE_LIBRTE_IEEE1588=n
 CONFIG_RTE_ETHDEV_QUEUE_STAT_CNTRS=16
 CONFIG_RTE_ETHDEV_RXTX_CALLBACKS=y
+CONFIG_RTE_ETHDEV_PROFILE_ITT_WASTED_RX_ITERATIONS=n
 
 #
 # Turn off Tx preparation stage
diff --git a/doc/guides/prog_guide/profile_app.rst 
b/doc/guides/prog_guide/profile_app.rst
index 54b546a..ca1c91f 100644
--- a/doc/guides/prog_guide/profile_app.rst
+++ b/doc/guides/prog_guide/profile_app.rst
@@ -39,7 +39,8 @@ Profiling on x86
 
 
 Intel processors provide performance counters to monitor events.
-Some tools provided by Intel, such as VTune, can be used to profile and 
benchmark an application.
+Some tools provided by Intel, such as Intel® VTune™ Amplifier, can be used
+to profile and benchmark an application.
 See the *VTune Performance Analyzer Essentials* publication from Intel Press 
for more information.
 
 For a DPDK application, this can be done in a Linux* application environment 
only.
@@ -59,6 +60,40 @@ Refer to the
 for details about application profiling.
 
 
+Empty cycles tracing
+
+
+Iterations that yielded no RX packets (empty cycles, wasted iterations) can
+be analyzed using VTune Amplifier. This profiling employs the
+`Instrumentation and Tracing Technology (ITT) API
+`_
+feature of VTune Amplifier and requires only reconfiguring the DPDK library,
+no changes in a DPDK application are needed.
+
+To trace wasted iterations on RX queues, first reconfigure DPDK with
+``CONFIG_RTE_ETHDEV_RXTX_CALLBACKS`` and
+``CONFIG_RTE_ETHDEV_PROFILE_ITT_WASTED_RX_ITERATIONS`` enabled.
+
+Then rebuild DPDK, specifying paths to the ITT header and library, which can
+be found in any VTune Amplifier distribution in the *include* and *lib*
+directories respectively:
+
+.. code-block:: console
+
+make EXTRA_CFLAGS=-I \
+ EXTRA_LDLIBS="-L -littnotify"
+
+Finally, to see wasted iterations in your performance analysis results,
+select the *"Analyze user tasks, events, and counters"* checkbox in the
+*"Analysis Type"* tab when configuring analysis via VTune Amplifier GUI.
+Alternatively, when running VTune Amplifier via command line, specify
+``-knob enable-user-tasks=true`` option.
+
+Collected regions of wasted iterations will be marked on VTune Amplifier's
+timeline as ITT tasks. These ITT tasks have predefined names, containing
+Ethernet device and RX queue identifiers.
+
+
 Profiling on ARM64
 --
 
diff --git a/lib/librte_ether/Makefile b/lib/librte_ether/Makefile
index 27d9766..5f65d05 100644
--- a/lib/librte_ether/Makefile
+++ b/lib/librte_ether/Makefile
@@ -46,6 +46,7 @@ LIBABIVER := 6
 SRCS-y += rte_ethdev.c
 SRCS-y += rte_flow.c
 SRCS-y += rte_tm.c
+SRCS-y += ethdev_profile.c
 
 #
 # Export include files
diff --git a/lib/librte_ether/ethdev_profile.c 
b/lib/librte_ether/ethd

Re: [dpdk-dev] [PATCH] ethdev: revert use device name from device structure

[Adrien Mazarguil]

On Fri, Sep 22, 2017 at 12:30:07PM +0100, Ferruh Yigit wrote:
> This reverts commit a1e7c17555e8f77d520ba5f06ed26c00e77a2bd1.
> 
> Original commit assumes there is 1:1 mapping between physical device and
> ethdev port, so that device name can be used per port instead of ethdev
> name field.
> 
> But one physical device may have multiple ethdev ports and each port
> needs its own unique name.
> 
> One issue reported here:
> http://dpdk.org/ml/archives/users/2017-September/002484.html
> 
> So reverting back the commit to continue using ethdev name field per
> port.
> 
> Reported-by: Adrien Mazarguil 
> Signed-off-by: Ferruh Yigit 

Acked-by: Adrien Mazarguil 

-- 
Adrien Mazarguil
6WIND

[dpdk-dev] [RFC PATCH v2 0/1] eventtimer: introduce event timer adapter

[Erik Gabriel Carrillo]

Hi Jerin, et al.,

The following RFC patch contains enough of a skeleton implementation
for an event timer adapter to show how the API could call through an ops
structure to a plugin backend to provide different implementations.

In terms of differences from the original RFC, the most obvious one is
that I've replaced occurrences of "wheel" in the API with "adapter",
partly to reflect the similarity with the event_eth_rx_adapter, and
partly because "wheel" suggests an implementation that may not be
the one used (as in the case of the SW impl). 

The second big change is to replace API parameters specifying pointer
to rte_event_timer_adapter with ids, which seems more common throughout
DPDK.

Other changes include:
 - removed rte_event_timer_adapter_lookup() function, since APIs no longer
   accept pointer to rte_event_timer_adapter
 - Replaced RTE_EVENT_TIMER_SUCCESS_{ARM,CANCEL} states with
   RTE_EVENT_TIMER_ARMED, since the purpose of the SUCCESS_CANCEL state was
   unclear

Please have a look and let me know what you think.

Thanks,
Gabriel


Erik Gabriel Carrillo (1):
  eventtimer: introduce event timer adapter

 drivers/event/sw/Makefile  |   1 +
 drivers/event/sw/sw_evdev.c|  11 +
 drivers/event/sw/sw_evdev_timer_adapter.c  | 124 +++
 lib/librte_eventdev/Makefile   |   3 +
 lib/librte_eventdev/rte_event_timer_adapter.c  | 227 
 lib/librte_eventdev/rte_event_timer_adapter.h  | 388 +
 .../rte_event_timer_adapter_driver.h   | 128 +++
 lib/librte_eventdev/rte_eventdev.h |   2 +-
 lib/librte_eventdev/rte_eventdev_pmd.h |  15 +
 lib/librte_eventdev/rte_eventdev_version.map   |  14 +
 test/test/Makefile |   1 +
 test/test/test_event_timer_adapter.c   | 183 ++
 12 files changed, 1096 insertions(+), 1 deletion(-)
 create mode 100644 drivers/event/sw/sw_evdev_timer_adapter.c
 create mode 100644 lib/librte_eventdev/rte_event_timer_adapter.c
 create mode 100644 lib/librte_eventdev/rte_event_timer_adapter.h
 create mode 100644 lib/librte_eventdev/rte_event_timer_adapter_driver.h
 create mode 100644 test/test/test_event_timer_adapter.c

-- 
2.6.4

[dpdk-dev] [RFC PATCH v2 1/1] eventtimer: introduce event timer adapter

[Erik Gabriel Carrillo]

Signed-off-by: Erik Gabriel Carrillo 
---
v2
- Added ops structure and stubbed out plugin for SW impl
- Added unit test stubs
- Replaced occurrences of "wheel" in API with "adapter"
- Replaced occurrences of pointer to struct rte_event_timer_adapter with ids
- Removed rte_event_timer_adapter_lookup() function
- Replaced RTE_EVENT_TIMER_SUCCESS_{ARM,CANCEL} states with
  RTE_EVENT_TIMER_ARMED

 drivers/event/sw/Makefile  |   1 +
 drivers/event/sw/sw_evdev.c|  11 +
 drivers/event/sw/sw_evdev_timer_adapter.c  | 124 +++
 lib/librte_eventdev/Makefile   |   3 +
 lib/librte_eventdev/rte_event_timer_adapter.c  | 227 
 lib/librte_eventdev/rte_event_timer_adapter.h  | 388 +
 .../rte_event_timer_adapter_driver.h   | 128 +++
 lib/librte_eventdev/rte_eventdev.h |   2 +-
 lib/librte_eventdev/rte_eventdev_pmd.h |  15 +
 lib/librte_eventdev/rte_eventdev_version.map   |  14 +
 test/test/Makefile |   1 +
 test/test/test_event_timer_adapter.c   | 183 ++
 12 files changed, 1096 insertions(+), 1 deletion(-)
 create mode 100644 drivers/event/sw/sw_evdev_timer_adapter.c
 create mode 100644 lib/librte_eventdev/rte_event_timer_adapter.c
 create mode 100644 lib/librte_eventdev/rte_event_timer_adapter.h
 create mode 100644 lib/librte_eventdev/rte_event_timer_adapter_driver.h
 create mode 100644 test/test/test_event_timer_adapter.c

diff --git a/drivers/event/sw/Makefile b/drivers/event/sw/Makefile
index 857a87c..f059915 100644
--- a/drivers/event/sw/Makefile
+++ b/drivers/event/sw/Makefile
@@ -55,6 +55,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV) += sw_evdev.c
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV) += sw_evdev_worker.c
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV) += sw_evdev_scheduler.c
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV) += sw_evdev_xstats.c
+SRCS-$(CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV) += sw_evdev_timer_adapter.c
 
 # export include files
 SYMLINK-y-include +=
diff --git a/drivers/event/sw/sw_evdev.c b/drivers/event/sw/sw_evdev.c
index 9c534b7..bbbf80a 100644
--- a/drivers/event/sw/sw_evdev.c
+++ b/drivers/event/sw/sw_evdev.c
@@ -49,6 +49,8 @@
 #define SCHED_QUANTA_ARG "sched_quanta"
 #define CREDIT_QUANTA_ARG "credit_quanta"
 
+extern const struct rte_event_timer_adapter_ops sw_event_adapter_timer_ops;
+
 static void
 sw_info_get(struct rte_eventdev *dev, struct rte_event_dev_info *info);
 
@@ -733,6 +735,13 @@ static int32_t sw_sched_service_func(void *args)
 }
 
 static int
+sw_event_timer_adapter_ctrl(void *arg)
+{
+   *(const void **)arg = &sw_event_adapter_timer_ops;
+   return 0;
+}
+
+static int
 sw_probe(struct rte_vdev_device *vdev)
 {
static const struct rte_eventdev_ops evdev_sw_ops = {
@@ -756,6 +765,8 @@ sw_probe(struct rte_vdev_device *vdev)
.xstats_get_names = sw_xstats_get_names,
.xstats_get_by_name = sw_xstats_get_by_name,
.xstats_reset = sw_xstats_reset,
+
+   .event_timer_adapter_ctrl = sw_event_timer_adapter_ctrl,
};
 
static const char *const args[] = {
diff --git a/drivers/event/sw/sw_evdev_timer_adapter.c 
b/drivers/event/sw/sw_evdev_timer_adapter.c
new file mode 100644
index 000..5edab0a
--- /dev/null
+++ b/drivers/event/sw/sw_evdev_timer_adapter.c
@@ -0,0 +1,124 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2016-2017 Intel Corporation. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRI

[dpdk-dev] [PATCH v3] net/virtio: fix of untrusted scalar value

[Daniel Mrzyglod]

The unscrutinized value may be incorrectly assumed to be within a certain
range by later operations.

In vhost_user_read: An unscrutinized value from an untrusted source used
in a trusted context - the value of sz_payload may be harmfull and we need
limit them to the max value of payload.

Coverity issue: 139601

Fixes: 6a84c37e3975 ("net/virtio-user: add vhost-user adapter layer")
Cc: jianfeng@intel.com
Cc: sta...@dpdk.org

Signed-off-by: Daniel Mrzyglod 
---
v3:
* there were wrong v2 email adress for stable dpdk mailinglist
* fix compilation errors

v2:
* Add Cc for stable in gitlog massage
* Add Coverity line
* v1 was acked by Acked-by: Jianfeng Tan 


 drivers/net/virtio/virtio_user/vhost_user.c | 4 
 1 file changed, 4 insertions(+)

diff --git a/drivers/net/virtio/virtio_user/vhost_user.c 
b/drivers/net/virtio/virtio_user/vhost_user.c
index 4ad7b21..97bd832 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -130,6 +130,10 @@ vhost_user_read(int fd, struct vhost_user_msg *msg)
}
 
sz_payload = msg->size;
+
+   if ((size_t)sz_payload > sizeof(msg->payload))
+   goto fail;
+
if (sz_payload) {
ret = recv(fd, (void *)((char *)msg + sz_hdr), sz_payload, 0);
if (ret < sz_payload) {
-- 
2.7.4

Re: [dpdk-dev] [PATCH] hash: optimize the softrss computation

[De Lara Guarch, Pablo]

Hi,

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Vladimir
> Medvedkin
> Sent: Tuesday, August 22, 2017 4:57 PM
> To: Yangchao Zhou 
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH] hash: optimize the softrss computation
> 
> Hi,
> 
> 2017-08-22 15:02 GMT+03:00 Yangchao Zhou :
> 
> > Use rte_bsf32 and fast bit unset operation to optimize the softrss
> > computation.
> > The following measurements shows improvement over the default softrss
> > computation function.
> >
> > tuple lens old(cycles) new(cycles)
> > 31225 337
> > 93743 992
> >
> > Signed-off-by: Yangchao Zhou 

Commit message is a bit long. Could you wrap it to have less than 75 characters 
per line?
Although, the patch itself has lines with more than 90 characters, which should 
be avoided:

http://dpdk.org/ml/archives/test-report/2017-August/027283.html

Last thing, the "Reviewed-by" has an extra space, before the name.

Thanks,
Pablo

> >
> Looks good for me. Thanks!
> 
> Reviewed-by:  Medvedkin Vladimir 
> 
> --
> Regards,
> Vladimir

Re: [dpdk-dev] [PATCH] [PATCH v3] lib/librte_sched: fix update tc_credits

[Dumitrescu, Cristian]

> 
> Signed-off-by: Olivier Chirossel 
> ---
>  doc/guides/prog_guide/qos_framework.rst |  10 ++-
>  lib/librte_sched/rte_sched.c| 124
> ++--
>  2 files changed, 112 insertions(+), 22 deletions(-)
> 
> diff --git a/doc/guides/prog_guide/qos_framework.rst
> b/doc/guides/prog_guide/qos_framework.rst
> index f3f60b8..e566ff9 100644
> --- a/doc/guides/prog_guide/qos_framework.rst
> +++ b/doc/guides/prog_guide/qos_framework.rst
> @@ -799,6 +799,9 @@ as described in :numref:`table_qos_10` and
> :numref:`table_qos_11`.
> | 4 | tc_credits| Bytes | Current upper limit for the
> number of credits that can be consumed by |
> |   |   |   | the current traffic class for
> the remainder of the current|
> |   |   |   | enforcement period.
>|
> +   |   |   |   | when The credits is update
> (every tc_period) the  |
> +   |   |   |   | tc_credits_per_period is added
> to the value (tc_credits) if the new   |
> +   |   |   |   | value is lower than tc_rate.
> else the value is set to tc_rate.|
> |   |   |   |
>|
> 
> +---+---+---+--
> -+
>  @@ -819,8 +822,11 @@ as described in :numref:`table_qos_10` and
> :numref:`table_qos_11`.
> |   |  |
>|
> |   |  | if (time >= tc_time) {
>|
> |   |  |
>|
> -   |   |  | tc_credits = tc_credits_per_period;
>|
> -   |   |  |
>|
> +   |   |  | if (tc_credits +
> tc_credits_per_period < tc_rate) {|
> +   |   |  |   tc_credits +=
> tc_credits_per_period |
> +   |   |  | else {
>|
> +   |   |  |   tc_credits = tc_rate
>|
> +   |   |  | }
>|
> |   |  | tc_time = time + tc_period;
>|
> |   |  |
>|
> |   |  | }
>|
> diff --git a/lib/librte_sched/rte_sched.c b/lib/librte_sched/rte_sched.c
> index b7cba11..732d5ef 100644
> --- a/lib/librte_sched/rte_sched.c
> +++ b/lib/librte_sched/rte_sched.c
> @@ -85,6 +85,7 @@ struct rte_sched_subport {
>   /* Traffic classes (TCs) */
>   uint64_t tc_time; /* time of next update */
> + uint32_t tc_rate[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
>   uint32_t
> tc_credits_per_period[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
>   uint32_t tc_credits[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
>   uint32_t tc_period;
> @@ -109,6 +110,7 @@ struct rte_sched_pipe_profile {
>   uint32_t tb_size;
>   /* Pipe traffic classes */
> + uint32_t tc_rate[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
>   uint32_t tc_period;
>   uint32_t
> tc_credits_per_period[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
>   uint8_t tc_ov_weight;
> @@ -568,11 +570,14 @@ rte_sched_port_config_pipe_profile_table(struct
> rte_sched_port *port, struct rte
>   /* Traffic Classes */
>   dst->tc_period = rte_sched_time_ms_to_bytes(src-
> >tc_period,
>   params->rate);
> -
> - for (j = 0; j < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; j++)
> +
> + for (j = 0; j < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; j++) {
> + dst->tc_rate[j] = src->tc_rate[j];
>   dst->tc_credits_per_period[j]
>   = rte_sched_time_ms_to_bytes(src-
> >tc_period,
>src->tc_rate[j]);
> + }
> +
>   #ifdef RTE_SCHED_SUBPORT_TC_OV
>   dst->tc_ov_weight = src->tc_ov_weight;
> @@ -838,6 +843,7 @@ rte_sched_subport_config(struct rte_sched_port
> *port,
>   /* Traffic Classes (TCs) */
>   s->tc_period = rte_sched_time_ms_to_bytes(params->tc_period,
> port->rate);
>   for (i = 0; i < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; i++) {
> + s->tc_rate[i] =  params->tc_rate[i];
>   s->tc_credits_per_period[i]
>   = rte_sched_time_ms_to_bytes(params->tc_period,
>

[dpdk-dev] [PATCH v3 2/7] app/crypto-perf: parse AEAD data from vectors

[Pablo de Lara]

Since DPDK 17.08, there is specific parameters
for AEAD algorithm, like AES-GCM. When verifying
crypto operations with test vectors, the parser
was not reading AEAD data (such as IV or key).

Fixes: 8a5b494a7f99 ("app/test-crypto-perf: add AEAD parameters")
Cc: sta...@dpdk.org

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_test_vector_parsing.c | 55 
 1 file changed, 55 insertions(+)

diff --git a/app/test-crypto-perf/cperf_test_vector_parsing.c 
b/app/test-crypto-perf/cperf_test_vector_parsing.c
index 148a604..3952632 100644
--- a/app/test-crypto-perf/cperf_test_vector_parsing.c
+++ b/app/test-crypto-perf/cperf_test_vector_parsing.c
@@ -116,6 +116,20 @@ show_test_vector(struct cperf_test_vector *test_vector)
printf("\n");
}
 
+   if (test_vector->aead_key.data) {
+   printf("\naead_key =\n");
+   for (i = 0; i < test_vector->aead_key.length; ++i) {
+   if ((i % wrap == 0) && (i != 0))
+   printf("\n");
+   if (i == (uint32_t)(test_vector->aead_key.length - 1))
+   printf("0x%02x", test_vector->aead_key.data[i]);
+   else
+   printf("0x%02x, ",
+   test_vector->aead_key.data[i]);
+   }
+   printf("\n");
+   }
+
if (test_vector->cipher_iv.data) {
printf("\ncipher_iv =\n");
for (i = 0; i < test_vector->cipher_iv.length; ++i) {
@@ -142,6 +156,19 @@ show_test_vector(struct cperf_test_vector *test_vector)
printf("\n");
}
 
+   if (test_vector->aead_iv.data) {
+   printf("\naead_iv =\n");
+   for (i = 0; i < test_vector->aead_iv.length; ++i) {
+   if ((i % wrap == 0) && (i != 0))
+   printf("\n");
+   if (i == (uint32_t)(test_vector->aead_iv.length - 1))
+   printf("0x%02x", test_vector->aead_iv.data[i]);
+   else
+   printf("0x%02x, ", 
test_vector->aead_iv.data[i]);
+   }
+   printf("\n");
+   }
+
if (test_vector->ciphertext.data) {
printf("\nciphertext =\n");
for (i = 0; i < test_vector->ciphertext.length; ++i) {
@@ -345,6 +372,20 @@ parse_entry(char *entry, struct cperf_test_vector *vector,
vector->auth_key.length = opts->auth_key_sz;
}
 
+   } else if (strstr(key_token, "aead_key")) {
+   rte_free(vector->aead_key.data);
+   vector->aead_key.data = data;
+   if (tc_found)
+   vector->aead_key.length = data_length;
+   else {
+   if (opts->aead_key_sz > data_length) {
+   printf("Global aead_key shorter than "
+   "aead_key_sz\n");
+   return -1;
+   }
+   vector->aead_key.length = opts->aead_key_sz;
+   }
+
} else if (strstr(key_token, "cipher_iv")) {
rte_free(vector->cipher_iv.data);
vector->cipher_iv.data = data;
@@ -373,6 +414,20 @@ parse_entry(char *entry, struct cperf_test_vector *vector,
vector->auth_iv.length = opts->auth_iv_sz;
}
 
+   } else if (strstr(key_token, "aead_iv")) {
+   rte_free(vector->aead_iv.data);
+   vector->aead_iv.data = data;
+   if (tc_found)
+   vector->aead_iv.length = data_length;
+   else {
+   if (opts->aead_iv_sz > data_length) {
+   printf("Global aead iv shorter than "
+   "aead_iv_sz\n");
+   return -1;
+   }
+   vector->aead_iv.length = opts->aead_iv_sz;
+   }
+
} else if (strstr(key_token, "ciphertext")) {
rte_free(vector->ciphertext.data);
vector->ciphertext.data = data;
-- 
2.9.4

[dpdk-dev] [PATCH v3 0/7] Crypto-perf app improvements

[Pablo de Lara]

This patchset includes some improvements in the Crypto performance
application, including app fixes and new parameter additions.

The last patch, in particular, introduces performance improvements.
Currently, crypto operations are allocated in a mempool and mbufs
in a different one. Then mbufs are extracted to an array, which is
looped through for all the crypto operations, impacting greatly
the performance, as much more memory is used.

Since crypto operations and mbufs are mapped 1:1, the can share the
same mempool object (similar to having the mbuf in the private data
of the crypto operation).
This improves performance, as it is only required to handle a single
mempool and the mbufs are obtained from the cache of the mempoool,
and not from an static array.

Changes in v3:
- Renamed "number of queue pairs" option from "--qps" to "--qp-nb",
  for more consistency

Changes in v2:

- Added support for multiple queue pairs
- Mempool for crypto operations and mbufs is now created
  using rte_mempool_create_empty(), rte_mempool_set_ops_byname(),
  rte_mempool_populate_default() and rte_mempool_obj_iter(),
  so mempool handler is set, as per Akhil's request.

Pablo de Lara (7):
  app/crypto-perf: set AAD after the crypto operation
  app/crypto-perf: parse AEAD data from vectors
  app/crypto-perf: parse segment size
  app/crypto-perf: overwrite mbuf when verifying
  app/crypto-perf: do not populate the mbufs at init
  app/crypto-perf: support multiple queue pairs
  app/crypto-perf: use single mempool

 app/test-crypto-perf/cperf_ops.c | 136 ++--
 app/test-crypto-perf/cperf_ops.h |   2 +-
 app/test-crypto-perf/cperf_options.h |   6 +-
 app/test-crypto-perf/cperf_options_parsing.c |  67 +++-
 app/test-crypto-perf/cperf_test_latency.c| 380 +++--
 app/test-crypto-perf/cperf_test_pmd_cyclecount.c | 382 +++---
 app/test-crypto-perf/cperf_test_throughput.c | 378 +++--
 app/test-crypto-perf/cperf_test_vector_parsing.c |  55 
 app/test-crypto-perf/cperf_test_verify.c | 399 ---
 app/test-crypto-perf/main.c  |  56 ++--
 doc/guides/tools/cryptoperf.rst  |  10 +-
 11 files changed, 1031 insertions(+), 840 deletions(-)

-- 
2.9.4

[dpdk-dev] [PATCH v3 1/7] app/crypto-perf: set AAD after the crypto operation

[Pablo de Lara]

Instead of prepending the AAD (Additional Authenticated Data)
in the mbuf, it is easier to set after the crypto operation,
as it is a read-only value, like the IV, and then it is not
restricted to the size of the mbuf headroom.

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_ops.c | 16 
 app/test-crypto-perf/cperf_test_latency.c| 16 
 app/test-crypto-perf/cperf_test_pmd_cyclecount.c | 16 +++-
 app/test-crypto-perf/cperf_test_throughput.c | 15 +++
 app/test-crypto-perf/cperf_test_verify.c | 20 ++--
 5 files changed, 28 insertions(+), 55 deletions(-)

diff --git a/app/test-crypto-perf/cperf_ops.c b/app/test-crypto-perf/cperf_ops.c
index 88fb972..5be20d9 100644
--- a/app/test-crypto-perf/cperf_ops.c
+++ b/app/test-crypto-perf/cperf_ops.c
@@ -307,6 +307,8 @@ cperf_set_ops_aead(struct rte_crypto_op **ops,
uint16_t iv_offset)
 {
uint16_t i;
+   uint16_t aad_offset = iv_offset +
+   RTE_ALIGN_CEIL(test_vector->aead_iv.length, 16);
 
for (i = 0; i < nb_ops; i++) {
struct rte_crypto_sym_op *sym_op = ops[i]->sym;
@@ -318,11 +320,12 @@ cperf_set_ops_aead(struct rte_crypto_op **ops,
 
/* AEAD parameters */
sym_op->aead.data.length = options->test_buffer_size;
-   sym_op->aead.data.offset =
-   RTE_ALIGN_CEIL(options->aead_aad_sz, 16);
+   sym_op->aead.data.offset = 0;
 
-   sym_op->aead.aad.data = rte_pktmbuf_mtod(bufs_in[i], uint8_t *);
-   sym_op->aead.aad.phys_addr = rte_pktmbuf_mtophys(bufs_in[i]);
+   sym_op->aead.aad.data = rte_crypto_op_ctod_offset(ops[i],
+   uint8_t *, aad_offset);
+   sym_op->aead.aad.phys_addr = 
rte_crypto_op_ctophys_offset(ops[i],
+   aad_offset);
 
if (options->aead_op == RTE_CRYPTO_AEAD_OP_DECRYPT) {
sym_op->aead.digest.data = test_vector->digest.data;
@@ -360,6 +363,11 @@ cperf_set_ops_aead(struct rte_crypto_op **ops,
 
memcpy(iv_ptr, test_vector->aead_iv.data,
test_vector->aead_iv.length);
+
+   /* Copy AAD after the IV */
+   memcpy(ops[i]->sym->aead.aad.data,
+   test_vector->aad.data,
+   test_vector->aad.length);
}
}
 
diff --git a/app/test-crypto-perf/cperf_test_latency.c 
b/app/test-crypto-perf/cperf_test_latency.c
index 58b21ab..2a46af9 100644
--- a/app/test-crypto-perf/cperf_test_latency.c
+++ b/app/test-crypto-perf/cperf_test_latency.c
@@ -174,16 +174,6 @@ cperf_mbuf_create(struct rte_mempool *mempool,
goto error;
}
 
-   if (options->op_type == CPERF_AEAD) {
-   uint8_t *aead = (uint8_t *)rte_pktmbuf_prepend(mbuf,
-   RTE_ALIGN_CEIL(options->aead_aad_sz, 16));
-
-   if (aead == NULL)
-   goto error;
-
-   memcpy(aead, test_vector->aad.data, test_vector->aad.length);
-   }
-
return mbuf;
 error:
if (mbuf != NULL)
@@ -289,10 +279,12 @@ cperf_latency_test_constructor(struct rte_mempool 
*sess_mp,
snprintf(pool_name, sizeof(pool_name), "cperf_op_pool_cdev_%d",
dev_id);
 
-   uint16_t priv_size = sizeof(struct priv_op_data) +
+   uint16_t priv_size = RTE_ALIGN_CEIL(sizeof(struct priv_op_data) +
test_vector->cipher_iv.length +
test_vector->auth_iv.length +
-   test_vector->aead_iv.length;
+   test_vector->aead_iv.length, 16) +
+   RTE_ALIGN_CEIL(options->aead_aad_sz, 16);
+
ctx->crypto_op_pool = rte_crypto_op_pool_create(pool_name,
RTE_CRYPTO_OP_TYPE_SYMMETRIC, options->pool_sz,
512, priv_size, rte_socket_id());
diff --git a/app/test-crypto-perf/cperf_test_pmd_cyclecount.c 
b/app/test-crypto-perf/cperf_test_pmd_cyclecount.c
index 0c949f0..ef1aa7e 100644
--- a/app/test-crypto-perf/cperf_test_pmd_cyclecount.c
+++ b/app/test-crypto-perf/cperf_test_pmd_cyclecount.c
@@ -191,16 +191,6 @@ cperf_mbuf_create(struct rte_mempool *mempool, uint32_t 
segments_nb,
goto error;
}
 
-   if (options->op_type == CPERF_AEAD) {
-   uint8_t *aead = (uint8_t *)rte_pktmbuf_prepend(
-   mbuf, RTE_ALIGN_CEIL(options->aead_aad_sz, 16));
-
-   if (aead == NULL)
-   goto error;
-
-   memcpy(aead, test_vector->aad.data, test_vector->aad.length);
-   }
-
return mbuf;
 error:
if (mbuf != NULL)
@@ -301,9 +291,9 @@ cperf_pmd_cyclecount_

[dpdk-dev] [PATCH v3 3/7] app/crypto-perf: parse segment size

[Pablo de Lara]

Instead of parsing number of segments, from the command line,
parse segment size, as it is a more usual case to have
the segment size fixed and then different packet sizes
will require different number of segments.

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_ops.c | 24 
 app/test-crypto-perf/cperf_options.h |  4 +-
 app/test-crypto-perf/cperf_options_parsing.c | 38 
 app/test-crypto-perf/cperf_test_latency.c| 78 +++-
 app/test-crypto-perf/cperf_test_pmd_cyclecount.c | 78 +++-
 app/test-crypto-perf/cperf_test_throughput.c | 78 +++-
 app/test-crypto-perf/cperf_test_verify.c | 78 +++-
 doc/guides/tools/cryptoperf.rst  |  6 +-
 8 files changed, 249 insertions(+), 135 deletions(-)

diff --git a/app/test-crypto-perf/cperf_ops.c b/app/test-crypto-perf/cperf_ops.c
index 5be20d9..ad32065 100644
--- a/app/test-crypto-perf/cperf_ops.c
+++ b/app/test-crypto-perf/cperf_ops.c
@@ -175,6 +175,14 @@ cperf_set_ops_auth(struct rte_crypto_op **ops,
offset -= tbuf->data_len;
tbuf = tbuf->next;
}
+   /*
+* If there is not enough room in segment,
+* place the digest in the next segment
+*/
+   if ((tbuf->data_len - offset) < 
options->digest_sz) {
+   tbuf = tbuf->next;
+   offset = 0;
+   }
buf = tbuf;
}
 
@@ -256,6 +264,14 @@ cperf_set_ops_cipher_auth(struct rte_crypto_op **ops,
offset -= tbuf->data_len;
tbuf = tbuf->next;
}
+   /*
+* If there is not enough room in segment,
+* place the digest in the next segment
+*/
+   if ((tbuf->data_len - offset) < 
options->digest_sz) {
+   tbuf = tbuf->next;
+   offset = 0;
+   }
buf = tbuf;
}
 
@@ -346,6 +362,14 @@ cperf_set_ops_aead(struct rte_crypto_op **ops,
offset -= tbuf->data_len;
tbuf = tbuf->next;
}
+   /*
+* If there is not enough room in segment,
+* place the digest in the next segment
+*/
+   if ((tbuf->data_len - offset) < 
options->digest_sz) {
+   tbuf = tbuf->next;
+   offset = 0;
+   }
buf = tbuf;
}
 
diff --git a/app/test-crypto-perf/cperf_options.h 
b/app/test-crypto-perf/cperf_options.h
index 2f42cb6..6d339f4 100644
--- a/app/test-crypto-perf/cperf_options.h
+++ b/app/test-crypto-perf/cperf_options.h
@@ -11,7 +11,7 @@
 #define CPERF_TOTAL_OPS("total-ops")
 #define CPERF_BURST_SIZE   ("burst-sz")
 #define CPERF_BUFFER_SIZE  ("buffer-sz")
-#define CPERF_SEGMENTS_NB  ("segments-nb")
+#define CPERF_SEGMENT_SIZE ("segment-sz")
 #define CPERF_DESC_NB  ("desc-nb")
 
 #define CPERF_DEVTYPE  ("devtype")
@@ -71,7 +71,7 @@ struct cperf_options {
 
uint32_t pool_sz;
uint32_t total_ops;
-   uint32_t segments_nb;
+   uint32_t segment_sz;
uint32_t test_buffer_size;
uint32_t nb_descriptors;
 
diff --git a/app/test-crypto-perf/cperf_options_parsing.c 
b/app/test-crypto-perf/cperf_options_parsing.c
index f3508a4..d372691 100644
--- a/app/test-crypto-perf/cperf_options_parsing.c
+++ b/app/test-crypto-perf/cperf_options_parsing.c
@@ -328,17 +328,17 @@ parse_buffer_sz(struct cperf_options *opts, const char 
*arg)
 }
 
 static int
-parse_segments_nb(struct cperf_options *opts, const char *arg)
+parse_segment_sz(struct cperf_options *opts, const char *arg)
 {
-   int ret = parse_uint32_t(&opts->segments_nb, arg);
+   int ret = parse_uint32_t(&opts->segment_sz, arg);
 
if (ret) {
-   RTE_LOG(ERR, USER1, "failed to parse segments number\n");
+   RTE_LOG(ERR, USER1, "failed to parse segment size\n");
return -1;
}
 
-   if ((opts->segments_nb == 0) || (opts->segments_nb > 255)) {
-   RTE_LOG(ERR, USER1, "invali

[dpdk-dev] [PATCH v3 4/7] app/crypto-perf: overwrite mbuf when verifying

[Pablo de Lara]

When running the verify test, mbufs in the pool were
populated with the test vector loaded from a file.
To avoid limiting the number of operations to the pool size,
mbufs will be rewritten with the test vector, before
linking them to the crypto operations.

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_options_parsing.c |  7 --
 app/test-crypto-perf/cperf_test_verify.c | 35 
 2 files changed, 35 insertions(+), 7 deletions(-)

diff --git a/app/test-crypto-perf/cperf_options_parsing.c 
b/app/test-crypto-perf/cperf_options_parsing.c
index d372691..89f86a2 100644
--- a/app/test-crypto-perf/cperf_options_parsing.c
+++ b/app/test-crypto-perf/cperf_options_parsing.c
@@ -944,13 +944,6 @@ cperf_options_check(struct cperf_options *options)
}
 
if (options->test == CPERF_TEST_TYPE_VERIFY &&
-   options->total_ops > options->pool_sz) {
-   RTE_LOG(ERR, USER1, "Total number of ops must be less than or"
-   " equal to the pool size.\n");
-   return -EINVAL;
-   }
-
-   if (options->test == CPERF_TEST_TYPE_VERIFY &&
(options->inc_buffer_size != 0 ||
options->buffer_size_count > 1)) {
RTE_LOG(ERR, USER1, "Only one buffer size is allowed when "
diff --git a/app/test-crypto-perf/cperf_test_verify.c 
b/app/test-crypto-perf/cperf_test_verify.c
index ba9621b..dbfa661 100644
--- a/app/test-crypto-perf/cperf_test_verify.c
+++ b/app/test-crypto-perf/cperf_test_verify.c
@@ -187,6 +187,34 @@ cperf_mbuf_create(struct rte_mempool *mempool,
return NULL;
 }
 
+static void
+cperf_mbuf_set(struct rte_mbuf *mbuf,
+   const struct cperf_options *options,
+   const struct cperf_test_vector *test_vector)
+{
+   uint32_t segment_sz = options->segment_sz;
+   uint8_t *mbuf_data;
+   uint8_t *test_data =
+   (options->cipher_op == RTE_CRYPTO_CIPHER_OP_ENCRYPT) ?
+   test_vector->plaintext.data :
+   test_vector->ciphertext.data;
+   uint32_t remaining_bytes = options->max_buffer_size;
+
+   while (remaining_bytes) {
+   mbuf_data = rte_pktmbuf_mtod(mbuf, uint8_t *);
+
+   if (remaining_bytes <= segment_sz) {
+   memcpy(mbuf_data, test_data, remaining_bytes);
+   return;
+   }
+
+   memcpy(mbuf_data, test_data, segment_sz);
+   remaining_bytes -= segment_sz;
+   test_data += segment_sz;
+   mbuf = mbuf->next;
+   }
+}
+
 void *
 cperf_verify_test_constructor(struct rte_mempool *sess_mp,
uint8_t dev_id, uint16_t qp_id,
@@ -469,6 +497,13 @@ cperf_verify_test_runner(void *test_ctx)
ops_needed, ctx->sess, ctx->options,
ctx->test_vector, iv_offset);
 
+
+   /* Populate the mbuf with the test vector, for verification */
+   for (i = 0; i < ops_needed; i++)
+   cperf_mbuf_set(ops[i]->sym->m_src,
+   ctx->options,
+   ctx->test_vector);
+
 #ifdef CPERF_LINEARIZATION_ENABLE
if (linearize) {
/* PMD doesn't support scatter-gather and source buffer
-- 
2.9.4

[dpdk-dev] [PATCH v3 5/7] app/crypto-perf: do not populate the mbufs at init

[Pablo de Lara]

For throughput and latency tests, it is not required
to populate the mbufs with any test vector.
For verify test, there is already a function that rewrites
the mbufs every time they are going to be used with
crypto operations.

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_test_latency.c| 31 +++-
 app/test-crypto-perf/cperf_test_pmd_cyclecount.c | 31 +++-
 app/test-crypto-perf/cperf_test_throughput.c | 31 +++-
 app/test-crypto-perf/cperf_test_verify.c | 31 +++-
 4 files changed, 36 insertions(+), 88 deletions(-)

diff --git a/app/test-crypto-perf/cperf_test_latency.c 
b/app/test-crypto-perf/cperf_test_latency.c
index 7b9dc9f..acd8545 100644
--- a/app/test-crypto-perf/cperf_test_latency.c
+++ b/app/test-crypto-perf/cperf_test_latency.c
@@ -118,15 +118,10 @@ static struct rte_mbuf *
 cperf_mbuf_create(struct rte_mempool *mempool,
uint32_t segment_sz,
uint16_t segments_nb,
-   const struct cperf_options *options,
-   const struct cperf_test_vector *test_vector)
+   const struct cperf_options *options)
 {
struct rte_mbuf *mbuf;
uint8_t *mbuf_data;
-   uint8_t *test_data =
-   (options->cipher_op == RTE_CRYPTO_CIPHER_OP_ENCRYPT) ?
-   test_vector->plaintext.data :
-   test_vector->ciphertext.data;
uint32_t remaining_bytes = options->max_buffer_size;
 
mbuf = rte_pktmbuf_alloc(mempool);
@@ -137,15 +132,11 @@ cperf_mbuf_create(struct rte_mempool *mempool,
if (mbuf_data == NULL)
goto error;
 
-   if (options->max_buffer_size <= segment_sz) {
-   memcpy(mbuf_data, test_data, options->max_buffer_size);
-   test_data += options->max_buffer_size;
+   if (options->max_buffer_size <= segment_sz)
remaining_bytes = 0;
-   } else {
-   memcpy(mbuf_data, test_data, segment_sz);
-   test_data += segment_sz;
+   else
remaining_bytes -= segment_sz;
-   }
+
segments_nb--;
 
while (remaining_bytes) {
@@ -161,15 +152,11 @@ cperf_mbuf_create(struct rte_mempool *mempool,
if (mbuf_data == NULL)
goto error;
 
-   if (remaining_bytes <= segment_sz) {
-   memcpy(mbuf_data, test_data, remaining_bytes);
+   if (remaining_bytes <= segment_sz)
remaining_bytes = 0;
-   test_data += remaining_bytes;
-   } else {
-   memcpy(mbuf_data, test_data, segment_sz);
+   else
remaining_bytes -= segment_sz;
-   test_data += segment_sz;
-   }
+
segments_nb--;
}
 
@@ -257,7 +244,7 @@ cperf_latency_test_constructor(struct rte_mempool *sess_mp,
ctx->pkt_mbuf_pool_in,
options->segment_sz,
segments_nb,
-   options, test_vector);
+   options);
if (ctx->mbufs_in[mbuf_idx] == NULL)
goto err;
}
@@ -286,7 +273,7 @@ cperf_latency_test_constructor(struct rte_mempool *sess_mp,
if (options->out_of_place == 1) {
ctx->mbufs_out[mbuf_idx] = cperf_mbuf_create(
ctx->pkt_mbuf_pool_out, max_size,
-   1, options, test_vector);
+   1, options);
if (ctx->mbufs_out[mbuf_idx] == NULL)
goto err;
} else {
diff --git a/app/test-crypto-perf/cperf_test_pmd_cyclecount.c 
b/app/test-crypto-perf/cperf_test_pmd_cyclecount.c
index 872124f..962dc69 100644
--- a/app/test-crypto-perf/cperf_test_pmd_cyclecount.c
+++ b/app/test-crypto-perf/cperf_test_pmd_cyclecount.c
@@ -135,15 +135,10 @@ cperf_pmd_cyclecount_test_free(struct 
cperf_pmd_cyclecount_ctx *ctx,
 static struct rte_mbuf *
 cperf_mbuf_create(struct rte_mempool *mempool,
uint32_t segment_sz, uint16_t segments_nb,
-   const struct cperf_options *options,
-   const struct cperf_test_vector *test_vector)
+   const struct cperf_options *options)
 {
struct rte_mbuf *mbuf;
uint8_t *mbuf_data;
-   uint8_t *test_data =
-   (options->cipher_op == RTE_CRYPTO_CIPHER_OP_ENCRYPT) ?
-   test_vector->plaintext.data :
-   test_vector->ciphertext.data;
uint32_t remaining_bytes = options->max_buffer_size;
 
mbuf = rte_pktmbuf_alloc(mempool);
@@ -154,15 +149,11 @@ cperf_mbuf_create(struct rte_mempool *

[dpdk-dev] [PATCH v3 6/7] app/crypto-perf: support multiple queue pairs

[Pablo de Lara]

Add parameter "qps" in crypto performance app,
to create multiple queue pairs per device.

This new parameter is useful to have multiple logical
cores using a single crypto device, without needing
to initialize a crypto device per core.

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_options.h |  2 +
 app/test-crypto-perf/cperf_options_parsing.c | 22 ++
 app/test-crypto-perf/cperf_test_latency.c| 14 +++---
 app/test-crypto-perf/cperf_test_pmd_cyclecount.c |  7 +--
 app/test-crypto-perf/cperf_test_throughput.c | 14 +++---
 app/test-crypto-perf/cperf_test_verify.c | 14 +++---
 app/test-crypto-perf/main.c  | 56 ++--
 doc/guides/tools/cryptoperf.rst  |  4 ++
 8 files changed, 84 insertions(+), 49 deletions(-)

diff --git a/app/test-crypto-perf/cperf_options.h 
b/app/test-crypto-perf/cperf_options.h
index 6d339f4..468d5e2 100644
--- a/app/test-crypto-perf/cperf_options.h
+++ b/app/test-crypto-perf/cperf_options.h
@@ -15,6 +15,7 @@
 #define CPERF_DESC_NB  ("desc-nb")
 
 #define CPERF_DEVTYPE  ("devtype")
+#define CPERF_QP_NB("qp-nb")
 #define CPERF_OPTYPE   ("optype")
 #define CPERF_SESSIONLESS  ("sessionless")
 #define CPERF_OUT_OF_PLACE ("out-of-place")
@@ -74,6 +75,7 @@ struct cperf_options {
uint32_t segment_sz;
uint32_t test_buffer_size;
uint32_t nb_descriptors;
+   uint32_t nb_qps;
 
uint32_t sessionless:1;
uint32_t out_of_place:1;
diff --git a/app/test-crypto-perf/cperf_options_parsing.c 
b/app/test-crypto-perf/cperf_options_parsing.c
index 89f86a2..441cd61 100644
--- a/app/test-crypto-perf/cperf_options_parsing.c
+++ b/app/test-crypto-perf/cperf_options_parsing.c
@@ -364,6 +364,24 @@ parse_desc_nb(struct cperf_options *opts, const char *arg)
 }
 
 static int
+parse_qp_nb(struct cperf_options *opts, const char *arg)
+{
+   int ret = parse_uint32_t(&opts->nb_qps, arg);
+
+   if (ret) {
+   RTE_LOG(ERR, USER1, "failed to parse number of queue pairs\n");
+   return -1;
+   }
+
+   if ((opts->nb_qps == 0) || (opts->nb_qps > 256)) {
+   RTE_LOG(ERR, USER1, "invalid number of queue pairs 
specified\n");
+   return -1;
+   }
+
+   return 0;
+}
+
+static int
 parse_device_type(struct cperf_options *opts, const char *arg)
 {
if (strlen(arg) > (sizeof(opts->device_type) - 1))
@@ -680,6 +698,7 @@ static struct option lgopts[] = {
{ CPERF_BUFFER_SIZE, required_argument, 0, 0 },
{ CPERF_SEGMENT_SIZE, required_argument, 0, 0 },
{ CPERF_DESC_NB, required_argument, 0, 0 },
+   { CPERF_QP_NB, required_argument, 0, 0 },
 
{ CPERF_DEVTYPE, required_argument, 0, 0 },
{ CPERF_OPTYPE, required_argument, 0, 0 },
@@ -747,6 +766,7 @@ cperf_options_default(struct cperf_options *opts)
 
strncpy(opts->device_type, "crypto_aesni_mb",
sizeof(opts->device_type));
+   opts->nb_qps = 1;
 
opts->op_type = CPERF_CIPHER_THEN_AUTH;
 
@@ -789,6 +809,7 @@ cperf_opts_parse_long(int opt_idx, struct cperf_options 
*opts)
{ CPERF_BUFFER_SIZE,parse_buffer_sz },
{ CPERF_SEGMENT_SIZE,   parse_segment_sz },
{ CPERF_DESC_NB,parse_desc_nb },
+   { CPERF_QP_NB,  parse_qp_nb },
{ CPERF_DEVTYPE,parse_device_type },
{ CPERF_OPTYPE, parse_op_type },
{ CPERF_SESSIONLESS,parse_sessionless },
@@ -1032,6 +1053,7 @@ cperf_options_dump(struct cperf_options *opts)
printf("#\n");
printf("# cryptodev type: %s\n", opts->device_type);
printf("#\n");
+   printf("# number of queue pairs per device: %u\n", opts->nb_qps);
printf("# crypto operation: %s\n", cperf_op_type_strs[opts->op_type]);
printf("# sessionless: %s\n", opts->sessionless ? "yes" : "no");
printf("# out of place: %s\n", opts->out_of_place ? "yes" : "no");
diff --git a/app/test-crypto-perf/cperf_test_latency.c 
b/app/test-crypto-perf/cperf_test_latency.c
index acd8545..99b92d3 100644
--- a/app/test-crypto-perf/cperf_test_latency.c
+++ b/app/test-crypto-perf/cperf_test_latency.c
@@ -218,8 +218,8 @@ cperf_latency_test_constructor(struct rte_mempool *sess_mp,
if (ctx->sess == NULL)
goto err;
 
-   snprintf(pool_name, sizeof(pool_name), "cperf_pool_in_cdev_%d",
-   dev_id);
+   snprintf(pool_name, sizeof(pool_name), "cperf_pool_in_cdev_%d_qp_%d",
+   dev_id, qp_id);
 
uint32_t max_size = options->max_buffer_size + options->digest_sz;
uint16_t segments_nb = (max_size % options->segment_sz) ?
@@ -252,8 +252,8 @@ cperf_latency_test_constructor(struct rte_mempool *sess_mp,
if (options->out_of_place == 1) {
 
snprintf(pool_name, sizeof(poo

[dpdk-dev] [PATCH v3 7/7] app/crypto-perf: use single mempool

[Pablo de Lara]

In order to improve memory utilization, a single mempool
is created, containing the crypto operation and mbufs
(one if operation is in-place, two if out-of-place).
This way, a single object is allocated and freed
per operation, reducing the amount of memory in cache,
which improves scalability.

Signed-off-by: Pablo de Lara 
---
 app/test-crypto-perf/cperf_ops.c |  96 --
 app/test-crypto-perf/cperf_ops.h |   2 +-
 app/test-crypto-perf/cperf_test_latency.c| 361 +++---
 app/test-crypto-perf/cperf_test_pmd_cyclecount.c | 364 +++---
 app/test-crypto-perf/cperf_test_throughput.c | 358 +++---
 app/test-crypto-perf/cperf_test_verify.c | 367 +++
 6 files changed, 793 insertions(+), 755 deletions(-)

diff --git a/app/test-crypto-perf/cperf_ops.c b/app/test-crypto-perf/cperf_ops.c
index ad32065..f76dbdd 100644
--- a/app/test-crypto-perf/cperf_ops.c
+++ b/app/test-crypto-perf/cperf_ops.c
@@ -37,7 +37,7 @@
 
 static int
 cperf_set_ops_null_cipher(struct rte_crypto_op **ops,
-   struct rte_mbuf **bufs_in, struct rte_mbuf **bufs_out,
+   uint32_t src_buf_offset, uint32_t dst_buf_offset,
uint16_t nb_ops, struct rte_cryptodev_sym_session *sess,
const struct cperf_options *options,
const struct cperf_test_vector *test_vector __rte_unused,
@@ -48,10 +48,18 @@ cperf_set_ops_null_cipher(struct rte_crypto_op **ops,
for (i = 0; i < nb_ops; i++) {
struct rte_crypto_sym_op *sym_op = ops[i]->sym;
 
+   ops[i]->status = RTE_CRYPTO_OP_STATUS_NOT_PROCESSED;
rte_crypto_op_attach_sym_session(ops[i], sess);
 
-   sym_op->m_src = bufs_in[i];
-   sym_op->m_dst = bufs_out[i];
+   sym_op->m_src = (struct rte_mbuf *)((uint8_t *)ops[i] +
+   src_buf_offset);
+
+   /* Set dest mbuf to NULL if out-of-place (dst_buf_offset = 0) */
+   if (dst_buf_offset == 0)
+   sym_op->m_dst = NULL;
+   else
+   sym_op->m_dst = (struct rte_mbuf *)((uint8_t *)ops[i] +
+   dst_buf_offset);
 
/* cipher parameters */
sym_op->cipher.data.length = options->test_buffer_size;
@@ -63,7 +71,7 @@ cperf_set_ops_null_cipher(struct rte_crypto_op **ops,
 
 static int
 cperf_set_ops_null_auth(struct rte_crypto_op **ops,
-   struct rte_mbuf **bufs_in, struct rte_mbuf **bufs_out,
+   uint32_t src_buf_offset, uint32_t dst_buf_offset,
uint16_t nb_ops, struct rte_cryptodev_sym_session *sess,
const struct cperf_options *options,
const struct cperf_test_vector *test_vector __rte_unused,
@@ -74,10 +82,18 @@ cperf_set_ops_null_auth(struct rte_crypto_op **ops,
for (i = 0; i < nb_ops; i++) {
struct rte_crypto_sym_op *sym_op = ops[i]->sym;
 
+   ops[i]->status = RTE_CRYPTO_OP_STATUS_NOT_PROCESSED;
rte_crypto_op_attach_sym_session(ops[i], sess);
 
-   sym_op->m_src = bufs_in[i];
-   sym_op->m_dst = bufs_out[i];
+   sym_op->m_src = (struct rte_mbuf *)((uint8_t *)ops[i] +
+   src_buf_offset);
+
+   /* Set dest mbuf to NULL if out-of-place (dst_buf_offset = 0) */
+   if (dst_buf_offset == 0)
+   sym_op->m_dst = NULL;
+   else
+   sym_op->m_dst = (struct rte_mbuf *)((uint8_t *)ops[i] +
+   dst_buf_offset);
 
/* auth parameters */
sym_op->auth.data.length = options->test_buffer_size;
@@ -89,7 +105,7 @@ cperf_set_ops_null_auth(struct rte_crypto_op **ops,
 
 static int
 cperf_set_ops_cipher(struct rte_crypto_op **ops,
-   struct rte_mbuf **bufs_in, struct rte_mbuf **bufs_out,
+   uint32_t src_buf_offset, uint32_t dst_buf_offset,
uint16_t nb_ops, struct rte_cryptodev_sym_session *sess,
const struct cperf_options *options,
const struct cperf_test_vector *test_vector,
@@ -100,10 +116,18 @@ cperf_set_ops_cipher(struct rte_crypto_op **ops,
for (i = 0; i < nb_ops; i++) {
struct rte_crypto_sym_op *sym_op = ops[i]->sym;
 
+   ops[i]->status = RTE_CRYPTO_OP_STATUS_NOT_PROCESSED;
rte_crypto_op_attach_sym_session(ops[i], sess);
 
-   sym_op->m_src = bufs_in[i];
-   sym_op->m_dst = bufs_out[i];
+   sym_op->m_src = (struct rte_mbuf *)((uint8_t *)ops[i] +
+   src_buf_offset);
+
+   /* Set dest mbuf to NULL if out-of-place (dst_buf_offset 

[dpdk-dev] [PATCH v4 1/5] app/testpmd: add traffic management forwarding mode

[Jasvinder Singh]

This commit extends the testpmd application with new forwarding engine
that demonstrates the use of ethdev traffic management APIs and softnic
PMD for QoS traffic management.

In this mode, 5-level hierarchical tree of the QoS scheduler is built
with the help of ethdev TM APIs such as shaper profile add/delete,
shared shaper add/update, node add/delete, hierarchy commit, etc.
The hierarchical tree has following nodes; root node(x1, level 0),
subport node(x1, level 1), pipe node(x4096, level 2),
tc node(x16348, level 3), queue node(x65536, level 4).

During runtime, each received packet is first classified by mapping the
packet fields information to 5-tuples (HQoS subport, pipe, traffic class,
queue within traffic class, and color) and storing it in the packet mbuf
sched field. After classification, each packet is sent to softnic port
which prioritizes the transmission of the received packets, and
accordingly sends them on to the output interface.

To enable traffic management mode, following testpmd command is used;

$ ./testpmd -c c -n 4 --vdev
'net_softnic0,hard_name=:06:00.1,soft_tm=on' -- -i
--forward-mode=tm

This patchset has dependency on following patch series;
http://dpdk.org/ml/archives/dev/2017-September/075655.html
http://dpdk.org/ml/archives/dev/2017-September/075656.html
http://dpdk.org/ml/archives/dev/2017-September/075657.html
http://dpdk.org/ml/archives/dev/2017-September/075658.html


Signed-off-by: Jasvinder Singh 
---
v3 change:
- Implements feedback from Pablo[1]
  - add flag to check required librte_sched lib and softnic pmd
  - code cleanup

v2 change:
- change file name softnictm.c to tm.c
- change forward mode name to "tm"
- code clean up

[1] http://dpdk.org/ml/archives/dev/2017-September/075744.html

 app/test-pmd/Makefile  |   8 +
 app/test-pmd/testpmd.c |  15 +
 app/test-pmd/testpmd.h |  45 +++
 app/test-pmd/tm.c  | 863 +
 4 files changed, 931 insertions(+)
 create mode 100644 app/test-pmd/tm.c

diff --git a/app/test-pmd/Makefile b/app/test-pmd/Makefile
index c36be19..2c50f68 100644
--- a/app/test-pmd/Makefile
+++ b/app/test-pmd/Makefile
@@ -59,6 +59,10 @@ SRCS-y += csumonly.c
 SRCS-y += icmpecho.c
 SRCS-$(CONFIG_RTE_LIBRTE_IEEE1588) += ieee1588fwd.c
 
+ifeq ($(CONFIG_RTE_LIBRTE_PMD_SOFTNIC)$(CONFIG_RTE_LIBRTE_SCHED),yy)
+SRCS-y += tm.c
+endif
+
 ifeq ($(CONFIG_RTE_BUILD_SHARED_LIB),y)
 
 ifeq ($(CONFIG_RTE_LIBRTE_PMD_BOND),y)
@@ -81,6 +85,10 @@ ifeq ($(CONFIG_RTE_LIBRTE_PMD_XENVIRT),y)
 LDLIBS += -lrte_pmd_xenvirt
 endif
 
+ifeq ($(CONFIG_RTE_LIBRTE_PMD_SOFTNIC),y)
+LDLIBS += -lrte_pmd_softnic
+endif
+
 endif
 
 CFLAGS_cmdline.o := -D_GNU_SOURCE
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index e097ee0..d729267 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -167,6 +167,10 @@ struct fwd_engine * fwd_engines[] = {
&tx_only_engine,
&csum_fwd_engine,
&icmp_echo_engine,
+#if defined RTE_LIBRTE_PMD_SOFTNIC && defined RTE_LIBRTE_SCHED
+   &softnic_tm_engine,
+   &softnic_tm_bypass_engine,
+#endif
 #ifdef RTE_LIBRTE_IEEE1588
&ieee1588_fwd_engine,
 #endif
@@ -2085,6 +2089,17 @@ init_port_config(void)
(rte_eth_devices[pid].data->dev_flags &
 RTE_ETH_DEV_INTR_RMV))
port->dev_conf.intr_conf.rmv = 1;
+
+#if defined RTE_LIBRTE_PMD_SOFTNIC && defined RTE_LIBRTE_SCHED
+   /* Detect softnic port */
+   if (!strcmp(port->dev_info.driver_name, "net_softnic")) {
+   port->softnic_enable = 1;
+   memset(&port->softport, 0, sizeof(struct softnic_port));
+
+   if (!strcmp(cur_fwd_eng->fwd_mode_name, "tm"))
+   port->softport.tm_flag = 1;
+   }
+#endif
}
 }
 
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index 1d1ee75..d8cb679 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -84,6 +84,12 @@ typedef uint16_t streamid_t;
 
 #define MAX_QUEUE_ID ((1 << (sizeof(queueid_t) * 8)) - 1)
 
+#if defined RTE_LIBRTE_PMD_SOFTNIC && defined RTE_LIBRTE_SCHED
+#define TM_MODE1
+#else
+#define TM_MODE0
+#endif
+
 enum {
PORT_TOPOLOGY_PAIRED,
PORT_TOPOLOGY_CHAINED,
@@ -162,6 +168,37 @@ struct port_flow {
uint8_t data[]; /**< Storage for pattern/actions. */
 };
 
+#ifdef TM_MODE
+/**
+ * Soft port tm related parameters
+ */
+struct softnic_port_tm {
+   uint32_t hierarchy_config;  /**< flag for tm hierarchy config */
+
+   uint32_t n_subports_per_port;  /**< Num of subport nodes per port */
+   uint32_t n_pipes_per_subport;  /**< Num of pipe nodes per subport */
+
+   uint64_t tm_pktfield0_slabpos;  /**< Pkt field position for subport */
+   uint64_t tm_pktfield0_slabmask; /**< Pkt field mask for the subport */
+   uint64_t tm_pktfield0_slabshr;
+ 

[dpdk-dev] [PATCH v4 3/5] app/test-pmd: add CLI for shaper and wred profiles

[Jasvinder Singh]

Add following CLIs in testpmd application;
- commands to add/del shaper profile for TM hieraqrchy nodes.
- commands to add/update shared shapers
- commands to add/del WRED profiles for TM hiearchy leaf nodes.

Signed-off-by: Jasvinder Singh 
---
v4 change:
 - remove softnic specific checkes to make it generic

 app/test-pmd/cmdline.c |  34 ++
 app/test-pmd/cmdline_tm.c  | 741 +
 app/test-pmd/testpmd_cmdline.h |   7 +
 3 files changed, 782 insertions(+)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 769afe8..36a29ba 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -653,6 +653,33 @@ static void cmd_help_long_parsed(void *parsed_result,
"ptype mapping update (port_id) (hw_ptype) (sw_ptype)\n"
"Update a ptype mapping item on a port\n\n"
 
+   "add port tm node shaper profile (port_id) 
(shaper_profile_id)"
+   " (tb_rate) (tb_size)\n"
+   "   Add port tm node private shaper profile.\n\n"
+
+   "del port tm node shaper profile (port_id) 
(shaper_profile_id)\n"
+   "   Delete port tm node private shaper profile.\n\n"
+
+   "add port tm node shared shaper (port_id) 
(shared_shaper_id)"
+   " (shaper_profile_id)\n"
+   "   Add/update port tm node shared shaper.\n\n"
+
+   "del port tm node shared shaper (port_id) 
(shared_shaper_id)\n"
+   "   Delete port tm node shared shaper.\n\n"
+
+   "set port tm node shaper profile (port_id) (node_id)"
+   " (shaper_profile_id)\n"
+   "   Set port tm node shaper profile.\n\n"
+
+   "add port tm node wred profile (port_id) 
(wred_profile_id)"
+   " (color_g) (min_th_g) (max_th_g) (maxp_inv_g) 
(wq_log2_g)"
+   " (color_y) (min_th_y) (max_th_y) (maxp_inv_y) 
(wq_log2_y)"
+   " (color_r) (min_th_r) (max_th_r) (maxp_inv_r) 
(wq_log2_r)\n"
+   "   Add port tm node wred profile.\n\n"
+
+   "del port tm node wred profile (port_id) 
(wred_profile_id)\n"
+   "   Delete port tm node wred profile.\n\n"
+
, list_pkt_forwarding_modes()
);
}
@@ -14401,6 +14428,13 @@ cmdline_parse_ctx_t main_ctx[] = {
(cmdline_parse_inst_t *)&cmd_show_port_tm_node_cap,
(cmdline_parse_inst_t *)&cmd_show_port_tm_node_type,
(cmdline_parse_inst_t *)&cmd_show_port_tm_node_stats,
+   (cmdline_parse_inst_t *)&cmd_add_port_tm_node_shaper_profile,
+   (cmdline_parse_inst_t *)&cmd_del_port_tm_node_shaper_profile,
+   (cmdline_parse_inst_t *)&cmd_add_port_tm_node_shared_shaper,
+   (cmdline_parse_inst_t *)&cmd_del_port_tm_node_shared_shaper,
+   (cmdline_parse_inst_t *)&cmd_add_port_tm_node_wred_profile,
+   (cmdline_parse_inst_t *)&cmd_del_port_tm_node_wred_profile,
+   (cmdline_parse_inst_t *)&cmd_set_port_tm_node_shaper_profile,
NULL,
 };
 
diff --git a/app/test-pmd/cmdline_tm.c b/app/test-pmd/cmdline_tm.c
index 8139908..b875e75 100644
--- a/app/test-pmd/cmdline_tm.c
+++ b/app/test-pmd/cmdline_tm.c
@@ -652,3 +652,744 @@ cmdline_parse_inst_t cmd_show_port_tm_node_type = {
NULL,
},
 };
+
+/* *** Add Port TM Private Shaper Profile *** */
+struct cmd_add_port_tm_node_shaper_profile_result {
+   cmdline_fixed_string_t add;
+   cmdline_fixed_string_t port;
+   cmdline_fixed_string_t tm;
+   cmdline_fixed_string_t node;
+   cmdline_fixed_string_t shaper;
+   cmdline_fixed_string_t profile;
+   uint8_t port_id;
+   uint32_t shaper_id;
+   uint64_t tb_rate;
+   uint64_t tb_size;
+};
+
+cmdline_parse_token_string_t cmd_add_port_tm_node_shaper_profile_add =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_node_shaper_profile_result, add, "add");
+cmdline_parse_token_string_t cmd_add_port_tm_node_shaper_profile_port =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_node_shaper_profile_result,
+   port, "port");
+cmdline_parse_token_string_t cmd_add_port_tm_node_shaper_profile_tm =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_node_shaper_profile_result,
+   tm, "tm");
+cmdline_parse_token_string_t cmd_add_port_tm_node_shaper_profile_node =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_node_shaper_profile_result,
+   node, "node");
+cmdline_parse_token_string_t cmd_add_port_tm_node_shaper_profile_shaper =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_node_shaper_profile_result,
+   shaper, "

[dpdk-dev] [PATCH v4 2/5] app/test-pmd: add CLI for TM capability and stats

[Jasvinder Singh]

Add following CLIs to testpmd application;
- commands to display TM capability information.
  (per port, per hierarchy level and per hierarchy node)
- command to display hiearchy node type
- stats collection

Signed-off-by: Jasvinder Singh 
---
v4 change:
 - remove softnic specific checkes to make it generic

v3 change:
- Implements feedback from Pablo[1]
 - move TM API related CLIs into cmdline_tm.c 
 - split patch into small patches
 - replace link status check with testpmd fn port_is_started()

[1]http://dpdk.org/ml/archives/dev/2017-September/075748.html

 app/test-pmd/Makefile  |   1 +
 app/test-pmd/cmdline.c |  23 +-
 app/test-pmd/cmdline_tm.c  | 654 +
 app/test-pmd/testpmd_cmdline.h |  44 +++
 4 files changed, 721 insertions(+), 1 deletion(-)
 create mode 100644 app/test-pmd/cmdline_tm.c
 create mode 100755 app/test-pmd/testpmd_cmdline.h

diff --git a/app/test-pmd/Makefile b/app/test-pmd/Makefile
index 2c50f68..99dbe4d 100644
--- a/app/test-pmd/Makefile
+++ b/app/test-pmd/Makefile
@@ -59,6 +59,7 @@ SRCS-y += csumonly.c
 SRCS-y += icmpecho.c
 SRCS-$(CONFIG_RTE_LIBRTE_IEEE1588) += ieee1588fwd.c
 
+SRCS-y += cmdline_tm.c
 ifeq ($(CONFIG_RTE_LIBRTE_PMD_SOFTNIC)$(CONFIG_RTE_LIBRTE_SCHED),yy)
 SRCS-y += tm.c
 endif
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index ccdf239..769afe8 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -99,6 +99,7 @@
 #include 
 #endif
 #include "testpmd.h"
+#include "testpmd_cmdline.h"
 
 static struct cmdline *testpmd_cl;
 
@@ -230,6 +231,21 @@ static void cmd_help_long_parsed(void *parsed_result,
 
"clear vf stats (port_id) (vf_id)\n"
"Reset a VF's statistics.\n\n"
+
+   "show port tm cap (port_id)\n"
+   "   Display the port TM capability.\n\n"
+
+   "show port tm level cap (port_id) (level_id)\n"
+   "   Display the port TM hierarchical level 
capability.\n\n"
+
+   "show port tm node cap (port_id) (node_id)\n"
+   "   Display the port TM node capability.\n\n"
+
+   "show port tm node type (port_id) (node_id)\n"
+   "   Display the port TM node type.\n\n"
+
+   "show port tm node stats (port_id) (node_id) (clear)\n"
+   "   Display the port TM node stats.\n\n"
);
}
 
@@ -14189,7 +14205,7 @@ cmdline_parse_inst_t cmd_load_from_file = {
},
 };
 
-/* 

 */
+/* *** */
 
 /* list of instructions */
 cmdline_parse_ctx_t main_ctx[] = {
@@ -14380,6 +14396,11 @@ cmdline_parse_ctx_t main_ctx[] = {
(cmdline_parse_inst_t *)&cmd_ptype_mapping_replace,
(cmdline_parse_inst_t *)&cmd_ptype_mapping_reset,
(cmdline_parse_inst_t *)&cmd_ptype_mapping_update,
+   (cmdline_parse_inst_t *)&cmd_show_port_tm_cap,
+   (cmdline_parse_inst_t *)&cmd_show_port_tm_level_cap,
+   (cmdline_parse_inst_t *)&cmd_show_port_tm_node_cap,
+   (cmdline_parse_inst_t *)&cmd_show_port_tm_node_type,
+   (cmdline_parse_inst_t *)&cmd_show_port_tm_node_stats,
NULL,
 };
 
diff --git a/app/test-pmd/cmdline_tm.c b/app/test-pmd/cmdline_tm.c
new file mode 100644
index 000..8139908
--- /dev/null
+++ b/app/test-pmd/cmdline_tm.c
@@ -0,0 +1,654 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2017 Intel Corporation. All rights reserved.
+ *   All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *  

[dpdk-dev] [PATCH v4 4/5] app/test-pmd: add CLI for TM nodes and hierarchy commit

[Jasvinder Singh]

Add following CLIs in testpmd application;
- commands to add TM hierarchy nodes (leaf and nonleaf).
- command for runtime update of node weight.
- command to commit the TM hierarchy

Signed-off-by: Jasvinder Singh 
---
v4 change:
 - remove softnic specific checkes to make it generic

 app/test-pmd/cmdline.c |  25 +++
 app/test-pmd/cmdline_tm.c  | 497 +
 app/test-pmd/testpmd_cmdline.h |   5 +
 3 files changed, 527 insertions(+)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 36a29ba..de7f487 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -680,6 +680,26 @@ static void cmd_help_long_parsed(void *parsed_result,
"del port tm node wred profile (port_id) 
(wred_profile_id)\n"
"   Delete port tm node wred profile.\n\n"
 
+   "add port tm nonleaf node (port_id) (node_id) 
(parent_node_id)"
+   " (priority) (weight) (level_id) (shaper_profile_id)"
+   " (shared_shaper_id) (n_shared_shapers) 
(n_sp_priorities)\n"
+   "   Add port tm nonleaf node.\n\n"
+
+   "add port tm leaf node (port_id) (node_id) 
(parent_node_id)"
+   " (priority) (weight) (level_id) (cman_mode)"
+   " (wred_profile_id)\n"
+   "   Add port tm leaf node.\n\n"
+
+   "del port tm node (port_id) (node_id)\n"
+   "   Delete port tm node.\n\n"
+
+   "set port tm node parent (port_id) (node_id) 
(parent_node_id)"
+   " (priority) (weight)\n"
+   "   Set port tm node parent.\n\n"
+
+   "port tm hierarchy commit (port_id) (clean_on_fail)\n"
+   "   Commit tm hierarchy.\n\n"
+
, list_pkt_forwarding_modes()
);
}
@@ -14435,6 +14455,11 @@ cmdline_parse_ctx_t main_ctx[] = {
(cmdline_parse_inst_t *)&cmd_add_port_tm_node_wred_profile,
(cmdline_parse_inst_t *)&cmd_del_port_tm_node_wred_profile,
(cmdline_parse_inst_t *)&cmd_set_port_tm_node_shaper_profile,
+   (cmdline_parse_inst_t *)&cmd_add_port_tm_nonleaf_node,
+   (cmdline_parse_inst_t *)&cmd_add_port_tm_leaf_node,
+   (cmdline_parse_inst_t *)&cmd_del_port_tm_node,
+   (cmdline_parse_inst_t *)&cmd_set_port_tm_node_parent,
+   (cmdline_parse_inst_t *)&cmd_port_tm_hierarchy_commit,
NULL,
 };
 
diff --git a/app/test-pmd/cmdline_tm.c b/app/test-pmd/cmdline_tm.c
index b875e75..5fc946d 100644
--- a/app/test-pmd/cmdline_tm.c
+++ b/app/test-pmd/cmdline_tm.c
@@ -1393,3 +1393,500 @@ cmdline_parse_inst_t 
cmd_set_port_tm_node_shaper_profile = {
NULL,
},
 };
+
+/* *** Add Port TM nonleaf node *** */
+struct cmd_add_port_tm_nonleaf_node_result {
+   cmdline_fixed_string_t add;
+   cmdline_fixed_string_t port;
+   cmdline_fixed_string_t tm;
+   cmdline_fixed_string_t nonleaf;
+   cmdline_fixed_string_t node;
+   uint8_t port_id;
+   uint32_t node_id;
+   int32_t parent_node_id;
+   uint32_t priority;
+   uint32_t weight;
+   uint32_t level_id;
+   uint32_t shaper_profile_id;
+   uint32_t shared_shaper_id;
+   uint32_t n_shared_shapers;
+   uint32_t n_sp_priorities;
+};
+
+cmdline_parse_token_string_t cmd_add_port_tm_nonleaf_node_add =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_nonleaf_node_result, add, "add");
+cmdline_parse_token_string_t cmd_add_port_tm_nonleaf_node_port =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_nonleaf_node_result, port, "port");
+cmdline_parse_token_string_t cmd_add_port_tm_nonleaf_node_tm =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_nonleaf_node_result, tm, "tm");
+cmdline_parse_token_string_t cmd_add_port_tm_nonleaf_node_nonleaf =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_nonleaf_node_result, nonleaf, "nonleaf");
+cmdline_parse_token_string_t cmd_add_port_tm_nonleaf_node_node =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_add_port_tm_nonleaf_node_result, node, "node");
+cmdline_parse_token_num_t cmd_add_port_tm_nonleaf_node_port_id =
+   TOKEN_NUM_INITIALIZER(
+   struct cmd_add_port_tm_nonleaf_node_result,
+port_id, UINT8);
+cmdline_parse_token_num_t cmd_add_port_tm_nonleaf_node_node_id =
+   TOKEN_NUM_INITIALIZER(struct cmd_add_port_tm_nonleaf_node_result,
+node_id, UINT32);
+cmdline_parse_token_num_t cmd_add_port_tm_nonleaf_node_parent_node_id =
+   TOKEN_NUM_INITIALIZER(struct cmd_add_port_tm_nonleaf_node_result,
+parent_node_id, INT32);
+cmdline_parse_token_num_t cmd_add_port_tm_nonleaf_node_priority =
+   TOKEN_NUM_INITIALIZER(struct cmd_add_port_

[dpdk-dev] [PATCH v4 5/5] app/test-pmd: add CLI for TM packet classification

[Jasvinder Singh]

Add following CLIs in testpmd application;
- command to set the packet field mask and offset value for
  classification.
- command to set traffic class translation table entry

Signed-off-by: Jasvinder Singh 
---
 app/test-pmd/cmdline.c |  13 ++
 app/test-pmd/cmdline_tm.c  | 309 +
 app/test-pmd/testpmd_cmdline.h |   4 +
 3 files changed, 326 insertions(+)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index de7f487..cc0a505 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -700,6 +700,15 @@ static void cmd_help_long_parsed(void *parsed_result,
"port tm hierarchy commit (port_id) (clean_on_fail)\n"
"   Commit tm hierarchy.\n\n"
 
+#if defined RTE_LIBRTE_PMD_SOFTNIC && defined RTE_LIBRTE_SCHED
+   "set port tm pktfield (subport|pipe|tc) (port_id) 
offset"
+   " (offset) mask (mask)\n"
+   "   Set port tm packet field.\n\n"
+
+   "set port tm tc table (port_id) index (tb_index) tc 
(tc_id)"
+   " queue (queue id)\n"
+   "   Set port tm traffic class table entry.\n\n"
+#endif
, list_pkt_forwarding_modes()
);
}
@@ -14460,6 +14469,10 @@ cmdline_parse_ctx_t main_ctx[] = {
(cmdline_parse_inst_t *)&cmd_del_port_tm_node,
(cmdline_parse_inst_t *)&cmd_set_port_tm_node_parent,
(cmdline_parse_inst_t *)&cmd_port_tm_hierarchy_commit,
+#if defined RTE_LIBRTE_PMD_SOFTNIC && defined RTE_LIBRTE_SCHED
+   (cmdline_parse_inst_t *)&cmd_set_port_tm_pktfield,
+   (cmdline_parse_inst_t *)&cmd_set_port_tm_tc_table,
+#endif
NULL,
 };
 
diff --git a/app/test-pmd/cmdline_tm.c b/app/test-pmd/cmdline_tm.c
index 5fc946d..ddeb45b 100644
--- a/app/test-pmd/cmdline_tm.c
+++ b/app/test-pmd/cmdline_tm.c
@@ -1890,3 +1890,312 @@ cmdline_parse_inst_t cmd_port_tm_hierarchy_commit = {
NULL,
},
 };
+
+#if defined RTE_LIBRTE_PMD_SOFTNIC && defined RTE_LIBRTE_SCHED
+
+static int
+port_tm_pktfield_validate_mask(uint64_t mask, uint64_t n)
+{
+   int count = __builtin_popcountll(mask);
+   int pos_lead = sizeof(uint64_t) * 8 - __builtin_clzll(mask);
+   int pos_trail = __builtin_ctzll(mask);
+   int count_expected = __builtin_popcount(n - 1);
+
+   /* Handle the exceptions */
+   if (n == 0)
+   return -1;  /* Error */
+
+   if ((mask == 0) && (n == 1))
+   return 0;   /* OK */
+
+   if (((mask == 0) && (n != 1)) || ((mask != 0) && (n == 1)))
+   return -2;  /* Error */
+
+   /* Check that mask is contiguous */
+   if ((pos_lead - pos_trail) != count)
+   return -3;  /* Error */
+
+   /* Check that mask contains the expected number of bits set */
+   if (count != count_expected)
+   return -4;  /* Error */
+
+   return 0;   /* OK */
+   }
+
+/* *** Set Port TM Packet Fields *** */
+struct cmd_set_port_tm_pktfield_result {
+   cmdline_fixed_string_t set;
+   cmdline_fixed_string_t port;
+   cmdline_fixed_string_t tm;
+   cmdline_fixed_string_t pktfield;
+   cmdline_fixed_string_t type;
+   uint8_t port_id;
+   cmdline_fixed_string_t offset_string;
+   uint32_t offset;
+   cmdline_fixed_string_t mask_string;
+   uint64_t mask;
+};
+
+cmdline_parse_token_string_t cmd_set_port_tm_pktfield_set =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result, set, "set");
+cmdline_parse_token_string_t cmd_set_port_tm_pktfield_port =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result, port, "port");
+cmdline_parse_token_string_t cmd_set_port_tm_pktfield_tm =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result, tm, "tm");
+cmdline_parse_token_string_t cmd_set_port_tm_pktfield_pktfield =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result,
+   pktfield, "pktfield");
+cmdline_parse_token_string_t cmd_set_port_tm_pktfield_type =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result,
+   type, "subport#pipe#tc");
+cmdline_parse_token_num_t cmd_set_port_tm_pktfield_port_id =
+   TOKEN_NUM_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result,
+   port_id, UINT8);
+cmdline_parse_token_string_t cmd_set_port_tm_pktfield_offset_string =
+   TOKEN_STRING_INITIALIZER(
+   struct cmd_set_port_tm_pktfield_result,
+   offset_string, "offset");
+cmdline_parse_token_num_t cmd_set_port_tm_pktfield_offset =
+   TOKEN_NUM_INITIALIZER(
+   stru

Re: [dpdk-dev] [dpdk-stable] [PATCH] net/vmxnet3: fix dereference before null check

[Ferruh Yigit]

On 9/22/2017 1:39 PM, Michal Jastrzebski wrote:
> From: Tomasz Kulasek 
> 
> Coverity error:
> 
> check_after_deref: Null-checking rq suggests that it may be null, but it
>has already been dereferenced on all paths leading to
>the check.
> 
> This patch moves NULL checking of "rq" at the very beginning of the path
> before any dereference.
> 
> Coverity issue: 143468
> Fixes: 5aecdc17a97d ("vmxnet3: fix stop/restart")
> Cc: yongw...@vmware.com
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Tomasz Kulasek 
> ---
>  drivers/net/vmxnet3/vmxnet3_rxtx.c | 17 -
>  1 file changed, 8 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/net/vmxnet3/vmxnet3_rxtx.c 
> b/drivers/net/vmxnet3/vmxnet3_rxtx.c
> index d9cf437..4fcceb4 100644
> --- a/drivers/net/vmxnet3/vmxnet3_rxtx.c
> +++ b/drivers/net/vmxnet3/vmxnet3_rxtx.c
> @@ -259,17 +259,16 @@
>  {
>   int i;
>   vmxnet3_rx_queue_t *rq = rxq;
> - struct vmxnet3_hw *hw = rq->hw;
>   struct vmxnet3_cmd_ring *ring0, *ring1;
>   struct vmxnet3_comp_ring *comp_ring;
> - struct vmxnet3_rx_data_ring *data_ring = &rq->data_ring;
>   int size;
>  
> - if (rq != NULL) {

vmxnet3_dev_rx_queue_reset() is static function and only called from
single function [1], which already checks if the parameter is NULL.

What do you think just removing this check and keep rest same?

[1]
vmxnet3_dev_clear_queues()

> - /* Release both the cmd_rings mbufs */
> - for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
> - vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
> - }
> + if (rq == NULL)
> + return;
> +
> + /* Release both the cmd_rings mbufs */
> + for (i = 0; i < VMXNET3_RX_CMDRING_SIZE; i++)
> + vmxnet3_rx_cmd_ring_release_mbufs(&rq->cmd_ring[i]);
>  
>   ring0 = &rq->cmd_ring[0];
>   ring1 = &rq->cmd_ring[1];
> @@ -287,8 +286,8 @@
>  
>   size = sizeof(struct Vmxnet3_RxDesc) * (ring0->size + ring1->size);
>   size += sizeof(struct Vmxnet3_RxCompDesc) * comp_ring->size;
> - if (VMXNET3_VERSION_GE_3(hw) && rq->data_desc_size)
> - size += rq->data_desc_size * data_ring->size;
> + if (VMXNET3_VERSION_GE_3(rq->hw) && rq->data_desc_size)
> + size += rq->data_desc_size * rq->data_ring.size;
>  
>   memset(ring0->base, 0, size);
>  }
> 

Re: [dpdk-dev] [dpdk-stable] [PATCH] net/i40e: fix assignment of enum values

[Ferruh Yigit]

On 9/22/2017 1:36 PM, Michal Jastrzebski wrote:
> From: Tomasz Kulasek 
> 
> mixed_enums: Mixing enum types enum i40e_vsi_type and enum
>  virtchnl_vsi_type for type
> 
> Coverity issue 158651
> Fixes: a58860f68929 ("net/i40e/base: use new virtchnl header file")
> Cc: jingjing...@intel.com
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Tomasz Kulasek 
> ---
>  drivers/net/i40e/i40e_ethdev_vf.c | 10 +-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/i40e/i40e_ethdev_vf.c 
> b/drivers/net/i40e/i40e_ethdev_vf.c
> index 73c315a..a8d2740 100644
> --- a/drivers/net/i40e/i40e_ethdev_vf.c
> +++ b/drivers/net/i40e/i40e_ethdev_vf.c
> @@ -1295,7 +1295,15 @@ static int i40evf_dev_xstats_get(struct rte_eth_dev 
> *dev,
>   if (hw->mac.type == I40E_MAC_X722_VF)
>   vf->flags = I40E_FLAG_RSS_AQ_CAPABLE;
>   vf->vsi.vsi_id = vf->vsi_res->vsi_id;
> - vf->vsi.type = (enum i40e_vsi_type)vf->vsi_res->vsi_type;
> +
> + switch (vf->vsi_res->vsi_type) {
> + case VIRTCHNL_VSI_SRIOV:
> + vf->vsi.type = I40E_VSI_SRIOV;
> + break;
> + default:
> + vf->vsi.type = I40E_VSI_TYPE_UNKNOWN;

This changes the behavior.

Previously:
vsi_type == VIRTCHNL_VSI_TYPE_INVALID ?  type = I40E_VSI_MAIN

Now:
vsi_type == VIRTCHNL_VSI_TYPE_INVALID ?  type = I40E_VSI_TYPE_UNKNOWN

> + break;
> + }
>   vf->vsi.nb_qps = vf->vsi_res->num_queue_pairs;
>   vf->vsi.adapter = I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
>  
> 

Re: [dpdk-dev] [PATCH] net/enic: fix multi-process operation

[Ferruh Yigit]

On 9/11/2017 7:58 PM, John Daley wrote:
> - Use rte_malloc() instead of malloc() for the per device 'vdev' structure
>   so that it can be shared across processes.
> - Only initialize the device if the process type is RTE_PROC_PRIMARY
> - Only allow the primary process to do queue setup, start/stop, promisc
>   allmulticast, mac add/del, mtu.
> 
> Fixes: fefed3d1e62c ("enic: new driver")
> Cc: sta...@dpdk.org
> 
> Signed-off-by: John Daley 

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

Re: [dpdk-dev] [PATCH v8] ether: add support for vtune task tracing

[Thomas Monjalon]

22/09/2017 16:52, ilia.kura...@intel.com:
> From: Ilia Kurakin 
> 
> The patch simplifies DPDK applications analysis for developers which use
> Intel® VTune Amplifier.
> 
> The empty cycles are such iterations that yielded no RX packets. As far as
> DPDK is running in poll mode, wasting cycles is equal to wasting CPU time.
> Tracing such iterations can identify that device is underutilized. Tracing
> empty cycles becomes even more critical if a system uses a lot of Ethernet
> ports.
> 
> The patch gives possibility to analyze empty cycles without changing
> application code. All needs to be done is just to reconfigure and rebuild
> the DPDK itself with CONFIG_RTE_ETHDEV_PROFILE_ITT_WASTED_RX_ITERATIONS
> enbled. The important thing here is that this does not affect DPDK code.
> The profiling code is not being compiled if user does not specify config
> flag.
> 
> The patch provides common way to inject RX queues profiling and VTune
> specific implementation.
> 
> Signed-off-by: Ilia Kurakin 

Applied with previous Ack from Jerin, thanks

Re: [dpdk-dev] [PATCH v4 1/4] net/i40e: fix clear xstats bug in vf port

[Ferruh Yigit]

On 9/21/2017 7:32 AM, Wei Zhao wrote:
> There is a bug in vf clear xstats command, it do not
> record the statics data in offset struct member.So, vf
> need to keep record of xstats data from pf and update
> the statics according to offset.
> 
> Fixes: da61cd0849766 ("i40evf: add extended stats")
> 
> Signed-off-by: Wei Zhao 

Series Reviewed-by: Ferruh Yigit 

Re: [dpdk-dev] [PATCH v4 1/4] net/i40e: fix clear xstats bug in vf port

[Ferruh Yigit]

On 9/22/2017 6:13 PM, Ferruh Yigit wrote:
> On 9/21/2017 7:32 AM, Wei Zhao wrote:
>> There is a bug in vf clear xstats command, it do not
>> record the statics data in offset struct member.So, vf
>> need to keep record of xstats data from pf and update
>> the statics according to offset.
>>
>> Fixes: da61cd0849766 ("i40evf: add extended stats")
>>
>> Signed-off-by: Wei Zhao 
> 
> Series Reviewed-by: Ferruh Yigit 

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

Re: [dpdk-dev] [PATCH v4] net/af_packet: make bypass configurable

[Ferruh Yigit]

On 9/21/2017 3:43 PM, Ferruh Yigit wrote:
> On 9/21/2017 10:57 AM, Chas Williams wrote:
>> From: "Charles (Chas) Williams" 
>>
>> In certain situations, low speed interfaces, it may be desirable to
>> have the flow control provided by the kernel queueing disciplines.
>>
>> Signed-off-by: Chas Williams 
> 
> Reviewed-by: Ferruh Yigit 

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