guest_binding: Add dpdk driver binding options for guests

Adds uio_pci_generic and vfio_no_iommu options for guest
driver binding. In case of SR-IOV tests with guests attached
and uio_pci_generic is selected the option will be modified
to use igb_uio instead as uio_pci_generic is not supported.

JIRA: VSPERF-397

Change-Id: I56003addacc8bf0d024cce35d41b00dd0baa8cbc
signed-off-by: Christian Trautman <ctrautma@redhat.com>

Merge "ci: Enhance verify and merge jobs"

ci: Enhance verify and merge jobs

Verify and merge jobs will execute real vsperf TCs,
which won't require neither physical NICs nor traffic
generator. Script ./build-vsperf.sh will take care about
creation, update and activation of python virtual environment.
It will also install required system packages at nodes
running merge and verify jobs.
Additional improvements:
* set of simple sanity checks was added
* two new integration TCs were added; However test with
  VNF is not used now, because it requires a VM image.
* detailed logs are shown in case of failure

JIRA: VSPERF-391

Change-Id: I33e75e762fdc62656de92dc34f0ad7a484623603
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>

doc: fix copy-paste error in Moongen configuration section

JIRA: VSPERF-414

Change-Id: I1924c8a0c949695b4ec0699a206b9e39ee2ef724
Signed-off-by: Otto Sabart <osabart@redhat.com>

Merge "testpmd: Simplify GUEST's testpmd configuration"

testpmd: Simplify GUEST's testpmd configuration

Testpmd has a lot of parameters and vsperf was able
to modify only a few of them through dedicated GUEST's
specific configuration options. Old configuration
options were replaced by one generic GUEST_TESTPMD_PARAMS,
which allows to modify any of existing testpmd parameters.
Detection of testpmd execution was changed to be independent
on TXQ flags setting.
Documentation was modified to reflect GUEST_TESTPMD change.
Also doc formatting was fixed at several places.

JIRA: VSPERF-380

Change-Id: Ie0bf290320f3b22ef6be6b55ed553235bbcb4aff
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

licenses: Dedicated directory for non Apache files

All files, which do not comply to Apache 2.0 license,
are moved into dedicated 3rd_party directory. This will
simplify license checks performed on vsperf code.
It is not required to move l2fwd module from its
current location.

JIRA: VSPERF-413

Change-Id: I6b4bd6bb5542e6db9a7ef91ee3f5f2907f93e9b8
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

Merge "IETF Draft: Comments from IETF-96"

vsperf: launch should not depend on current $PWD

JIRA: VSPERF-392

Change-Id: I44866b7d3298617003b87b015eb08bf32c4014f7
Signed-off-by: Otto Sabart <osabart@redhat.com>

Merge "integration: Test vHost User numa awareness"

IETF Draft: Comments from IETF-96

JIRA: VSPERF-410

x Ramki on OVS features - the scope of benchmarking does not include
the ever-growing list of OVS features, only the general and
switch-agnostic features will be assessed.

x Justify the length of SOAK tests (with variability allowed),
mention that the goal is stability characterization, not
the typical short term benchmarks of performance.

x Fix the LTD Reference to point to Brahmaputra version
(which is frozen in the release docs)

Change-Id: I2f795cd42e7ec2e06e5cd35bbf0c1f336a5e43ca
Signed-off-by: Al Morton <acmorton@att.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Billy O'Mahony<billy.o.mahony@intel.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>

CLI: add Multiqueue params.

This patch implements the following new CLI parameters that
can be used in the --test-params list.

 1. guest_nic_queues, which overrides all GUEST_NIC_QUEUES values
 2. guest_testpmd_txq, which overrides all GUEST_TESTPMD_TXQ
 3. guest_testpmd_rxq, which overrides all GUEST_TESTPMD_RXQ
 4. guest_testpmd_nb_cores, which overrides all GUEST_TESTPMD_NB_CORES
    values
 5. guest_testpmd_cpu_mask, which overrides all GUEST_TESTPMD_CPU_MASK
    values
 6. vswitch_dpdk_multi_queues, which overrides VSWITCH_DPDK_MULTI_QUEUES
 7. guest_smp, which overrides all GUEST_SMP values
 8. guest_core_binding, which overrides all GUEST_CORE_BINDING values

When they are present in the CLI the corresponding settings from the
configuration files will be overridden.

As an example a usage can be:
./vsperf --test-params "guest_nic_queues=2;guest_testpmd_txq=1;\
guest_testpmd_rxq=1;guest_testpmd_nb_cores=1"...

v2: rebase + rework to add other CLI params as suggested and update doc.
v3: changes in 'testusage.rst' doc.

JIRA: VSPERF-371

Change-Id: I140d6e068c1653d21aaa0df27532a48c4a12c151
Signed-off-by: Antonio Fischetti <antonio.fischetti@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Martin Klozik <martinx.klozik@intel.com>

Merge "Integration Test: 4 serial VMs testcases."

integration: Test vHost User numa awareness

Open vSwitch with DPDK can optimize memory usage
in case of NUMA architecture to avoid unnecessary
memory access across NUMA slots. In a nutshell,
PMD threads serving virtual NICs are co-located at
the same NUMA slot as QEMU instance, which is using
these NICs.
This patch adds new (functional) integration testcase,
which verifies OVS vHost User numa awareness feature.
Step driven test objects were updated to allow a call
of OS utilies and evaluation of conditions. Also the
documentation was updated with the list of supported
test objects and their methods.

JIRA: VSPERF-377

Change-Id: I184e71e066d27b5b9fc9e6a9f7e240e2d1b5a0fa
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Ciara Loftus <ciara.loftus@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

Merge "pkt_gen: Spirent Testcenter RFC 2889 Support"

Merge "cmd_timeout: Add ovs command timeout configuration option"

cmd_timeout: Add ovs command timeout configuration option

Adds option for configurable timeout for ovs commands.

JIRA: VSPERF-396

Change-Id: Ie73dfea1a088a5c92bef705f1fe874ddeeaec5a7
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

bugfix: Correct VM handling by integration tests

Patch with multi VM support has changed the way
how data are copied into the shared folder. Also
new support of GUEST_ options expansion requires
proper initialization with exact number of VMs
involved in the test.
Code of class specific to integration tests
was updated to initialize both GUEST_ options
and VM shared dir correctly.
Original values of all GUEST_ options are restored
after the testcase execution. So other TCs can
use original values for configuration expansion
to required number of VMs.

JIRA: VSPERF-361

Change-Id: Ic1149fbc73f73a66982bb64173b2217e55597200
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

ci: Update default DPDK and OVS versions

VSPERF was updated to use DPDK16.07 and OVS 2.6.90
by default. This change should improve a performance.
It will also solve failures of VSPERF verify and merge
jobs at OPNFV Jenkins, because newer OVS version is
compatible with Linux Kernel 4.4.

JIRA: VSPERF-378

Change-Id: I636dd96facd3f6357f215290eb2382c9cc16150e
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

bugfix: Qemu shared dir failure

In case of testpmd or l2fwd loopback app, VSPERF copies
sources if these tools into qemu shared folder. Maximum
size of folder content is limited to approx. 500MB. This
limit is exceeded in case that DPDK is built for multiple
RTE_TARGETs. Unnecessary files are excluded from the
copy to spare size inside shared folder.

JIRA: VSPERF-398

Change-Id: Iab31a6ff2397f7031eae6a95d424317c3343d4d9
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

pkt_gen: Spirent Testcenter RFC 2889 Support

The changes/additions, apart from spirent testcenter-specific, also
includes
to some files in conf (01_testcase and 03_traffic) and core (result-
constants, traffic_controller and component_factory) folders.

Currently, only RFC2889 Forwarding testcase is supported.

Incorporated following review suggestions:
1. Log level fixes
2. Removing unused function in results_constants.py
3. Added documentation to docs/configguide/trafficgen.rst. Userguide
   will be updated once other RFC2889 tests are implemented.
4. string matching in component_factory.
5. Remove Trailing Whitespaces

JIRA: VSPERF-286

Change-Id: I0195720ab2f8cf2c3a5aa490d66166bdca0afcb0
Signed-off-by: Sridhar K. N. Rao <sridhar.rao@spirent.com>

Merge "paths: Support binary packages"

paths: Support binary packages

Currently VSPERF supports OVS, DPDK and QEMU built
from the source code only. In some cases it is required
to support installation of these tools from binary packages
available for given linux distribution. Thus VSPERF
configuration and code was modified to suport both source
and binary versions of tools. This can be configured perf
tool, so various combinations of source and binary version
are supported.
Together with new configuration also a handling of kernel
modules was modified to automatically detect and load module
dependencies.

JIRA: VSPERF-340
JIRA: VSPERF-339

Change-Id: I855cb438cbd8998bdc499613ea5e7de2526299d7
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Otto Sabart <osabart@redhat.com>

merge_buffers: Add option to disable merge buffer on qemu startup

In cases with VHost user and non-jumbo frame scenarios we can
disable merge buffers on the guest NICS to improve performance.
This patch adds this option to the 04_vnf conf file.

JIRA: VSPERF-388

Change-Id: Ic1f776e90fdccaa5d5a0c952c96811a8747fe936
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

bugfix: SRIOV used wrong device names

QemuPciPassthrough class responsible for SRIOV testing
used wrong device names for GUEST configuration. Details
of HOST interfaces connected to the traffic generator were
used to initialize forwarding applications inside the guest.
Code was fixed to use correct device configuration.
As part of this fix, old bug in VNF type detection was fixed.

JIRA: VSPERF-389

Change-Id: I269c9232012980c600c171571e90beb11e7b72cd
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

Integration Test: 4 serial VMs testcases.

Implementation of RFC2544 throughput/back2back and
continuous testcases with 4 serial VMs.

Topology is: Phy1-VM0-VM1-VM2-VM3-Phy2.

This patch requires Patch VSPERF-326 "Integration test
implementation: Multi-PVP with explicit Flows".

v2: rebased to master to fix conflicts.

JIRA: VSPERF-364

Change-Id: Ib731bf6dc4634d6347b1c521bb82326db90a6c63
Signed-off-by: Antonio Fischetti <antonio.fischetti@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Martin Klozik <martinx.klozik@intel.com>

pkt_gen: MoonGen code clean up

MoonGen code optimizations and general clean up.  No
functionality changes.

JIRA: VSPERF-351

Change-Id: Ida16f16eb218eb6f0a1c90594c38511bfbb2cf97
Signed-off-by: bmichalo <bmichalo@redhat.com>

Merge "docs: Fix links and formatting"

docs: Fix links and formatting

Documentation was updated as follows:
* section decorators were fixed in LTD
* links to testcase and deployment
  descriptions were fixed
* duplicate labels removed from LTP&LTD
* fixed typos in formatting and text
* links to scalability and stress tests
  fixed after their renaming

JIRA: VSPERF-349

Change-Id: I86c2b723231431388b69d3d696805de2a4e12435
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

Merge "multi VM: Fix p2p deployment"

multi VM: Fix p2p deployment

Revert enforced vsperf failure in case, that VnfController
is used in scenario without VNFs. Instance of VnfController
is created for all scenarios including p2p, so it must
be prepared for it.

JIRA: VSPERF-361

Change-Id: Icf6e88e7ea62044c09bb194c260ea878d6a197c9
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

Merge "test_spec: Remove formatted text file verions of IETF Draft"

Merge "multi VM: Multi VMs in serial or parallel"

docs: Remove links to Brahmaputra documentation

Hyperlinks to the yardstick documentation were updated
to refer to files generated from the master branch.

JIRA: VSPERF-349

Change-Id: I248795b6a951377885492515fa9b96a694c9da5d
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

Merge "vhost_affin_fixup: Change vhost thread from regex to pgrep usage"

test_spec: Remove formatted text file verions of IETF Draft

Remove files with IETF copyrights (Internet Draft .txt files).
I assume these files have to be removed in C-stable branch...

JIRA: VSPERF-387

Change-Id: I76709ab055f845cec0bec8d07b14c29dec349f54
Signed-off-by: Al Morton <acmorton@att.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

multi VM: Multi VMs in serial or parallel

Support for deployment scenarios with any number of VMs
in both serial and parallel configuration. Detailed
content of the patch:

* VswitchControllerPXP class for multi VM support
* pvvpxx and pvpvxx deployments for xx VMs in
  serial respective parallel configuration
* special GUEST_ options expansion to requested
  number of VMs;
* support of GUEST_ options specific macros
  #VMINDEX, #MAC(), #IP() and #EVAL()
* all GUEST specific options are turned to lists
  to be VM specific
* support for VM with 1 NIC
* support for VM with multiple NIC pairs; traffic
  is routed in serial or parallel between NIC paris
  based on deployment scenario
* support for PVVP and PVPV scenarios using VMs
  with different numbers of NICs

JIRA: VSPERF-361

Change-Id: I05bedbdfa9a81ea0166d9b03d83ae49d6cb8b19b
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>
Reviewed-by: Antonio Fischetti <antonio.fischetti@intel.com>

vhost_affin_fixup: Change vhost thread from regex to pgrep usage

Change vhost affinity process lookup to use pgrep instead of
ugly error prone regex method as suggested.

Change-Id: I8cac00a7f2d26ba1ce1815535a198f578159222c
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

pkt_gen: MoonGen Incorrect RX_MBPS Calculation

Misplaced parenthesis regarding THROUGHPUT_RX_MBPS

JIRA: VSPERF-381

Change-Id: Ie33cf6ca88164f0686a819715b5f9f7409072063
Signed-off-by: bmichalo <bmichalo@redhat.com>

docs: vloop-vnf changelog update

Documentation was updated to reflect recent changes
of vloop-vnf image.

JIRA: VSPERF-374

Change-Id: I65a34a7618593a7b77c38eb2fa4c87619b937346
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

Vanilla_Multi_Queue: Add vanilla ovs multi-queue functionality

Adds multi-queue for vanilla OVS using virtio-net.

TunTap ports will use multi_queue parameter when ports are
created/deleted if guest nic queues are enabled and
vswitch is ovs vanilla. Virtio net will now add guest nic
queues to qemu command line if appropriate.

Reworked multi-queue documentation to reflect these changes.

Adds vhost net thread affinitization which is recommended
when performing vanilla ovs multi-queue.

Guests will require ethtool if using l2fwd or linux bridge
as the loopback application when vanilla ovs multi-queue
is enabled.

Modified dpdk setting in vswitch conf to be better worded
for separation from vanilla ovs multi-queue.

Updated release and installation doc for new vloop image
that includes required ethtool utility.

JIRA: VSPERF-373

Change-Id: Idb550515190b1a93390308c11f54da368f962512
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

guest_drive_type: Adds guest drive configuration options

Adds boot and mounted drive configuration options.

JIRA: VSPERF-379

Change-Id: Ic0673a9b9f7a3430d7ac4db5fb9bf4228872f501
signed-off-by: Christian Trautman <ctrautma@redhat.com>

cuse: Remove vHost Cuse support

Support of vHost Cuse interface is not supported
by OVS anymore, so it was removed also from VSPERF.
Cuse can be used with older VSPERF release if needed.

JIRA: VSPERF-376

Change-Id: Id786bddf5fa9dce5706a93de7b089707a60317cd
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: Bill Michalowski <bmichalo@redhat.com>

Merge "old_style_multi-queue: Allows for multi-queue with ovs <= 2.5.0"

Integration Test: Multi-VM with parallel PVP connections.

Implementation of 3 testcases with multi-VMs in parallel.
Flows are explicitly set on UDP destination ports so that
each packet is sent to a specific VM.

Topology is: Phy1-VM1-Phy2, Phy1-VM2-Phy2,..

This patch requires change #16855 "ovs/ofctl: Fix validation
method for complex flows."

JIRA: VSPERF-326

Change-Id: I59d23354bc9c6db470a59faba13bdcd6f2beb19d
Signed-off-by: Antonio Fischetti <antonio.fischetti@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Martin Klozik <martinx.klozik@intel.com>

integration: Support for VxLAN TC without using overlay traffic gen.

The test case for verifying the vxlan tunneling feature without using any
ingress tunnel traffic. The virtual switch(OVS) is configured to mangle and
generate tunnel traffic in the deployment. The packet flow in the test case is
as follows

TRAFFIC-IN --> [ENCAP-PKT] --> [MOD-PKT] --> [DECAP-PKT] --> TRAFFIC-OUT

 ENCAP-PKT - Encapsulate the ingress packet to a tunnel type.
 MOD-PKT   - Modify the tunnel header to match the following decap interface.
 DECAP-PKT - Decapsulate the newly generated tunneled packet.

Change-Id: Ie24bacb3cb1c069bd60403e5a4ef8bcdf0e12e54
Signed-off-by: Sugesh Chandran <sugesh.chandran@intel.com>

old_style_multi-queue: Allows for multi-queue with ovs <= 2.5.0

Enables multi-queue option for OVS versions 2.5.0 or less with
dpdk. Instead of enabling MQ on a port by port basis this sets
the global MQ option after starting the ovs switch.

JIRA: VSPERF-372

Change-Id: I245c76779695518d0005122d8214bbf3f141bb96
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

docs: add vm2vm testing methodology

Change-Id: I8a89ee3eedd3b439f0243eaef2fafc50ee807739
Signed-off-by: Maryam Tahhan <maryam.tahhan@intel.com>

Merge "docs: add LTP to index.rst"

doc_fixup: Grammar and link fixes to documentation

Fixes minor grammar and link issues.

JIRA: VSPERF-349

Change-Id: Id8708aee5b099287b5c99e79ca6da9ab8778897f
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

pkt_gen: Spirent changes to Harmonize test/trial terminology

Added changes to make use of the tests parameter.
The change is minor.
Prior to this the value was taken from configuration file.
With this change, the tests value is used as passed.

JIRA: VSPERF-367

Change-Id: I9d1553d3b53dd19a79428ddaaacadb55fc27324f
Signed-off-by: Sridhar Rao <sridhar.rao@spirent.com>

docs: add LTP to index.rst

Fix build issues and add the LTP to index.rst.

JIRA: DOCS-95
Change-Id: I2d22646af935c544645c01600b92a5f51ede39ec
Signed-off-by: Maryam Tahhan <maryam.tahhan@intel.com>

pkt_gen: Update MoonGen Documention

Updating MoonGen configuration guide

JIRA: VSPERF-370

Change-Id: I577491a6ece249a6162e1a3595a8682b826e5734
Signed-off-by: bmichalo <bmichalo@redhat.com>

hugepage_fix: Fix free hugepage calcluation

Fixes hugepage free calculation when hugepage size is not
1GB. Calculation previously assumed 1GB in size. This change
allows for all sizes to be calculated properly.

Change-Id: Ib05800f51992e9757c412a66296a47ad29e88a3b
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

Merge "docs: split LTD into LTP and LTD."

Merge "integration_cleanup: Force testcase cleanup to always run"

Merge "Xena_2544_tput_options: Add options for rfc 2544 testing"

Merge "hugepages: change default num pages + deallocate"

docs: split LTD into LTP and LTD.

Change-Id: Id0b2ce7460e76ffb3d478bd255a7e5b8e75ba001
Signed-off-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>
Reviewed-by: <sridhar.rao@spirent.com>

integration_cleanup: Force testcase cleanup to always run

During testcase development on the integration side, if
namespaces are used or veth ports and an exception is raised.
The namespaces and veth ports are not properly cleanup up
because the run_finalize does not get executed.

This patch wraps the testing block in a try block where
a finally will ALWAYS execute the run_finalize thus
preventing the possiblity of leftover testing artifacts.

JIRA: VSPERF-328

Change-Id: I4ea5f29a48b9171f330847f34b5b30ef3ab22f27
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

Xena_2544_tput_options: Add options for rfc 2544 testing

Adds options to Xena for testing options when running RFC 2544
throughput testing.

JIRA: VSPERF-369

Change-Id: I3145bd1a4d697d7b06cafda4f9d77cce859d1839
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

Merge "test_spec: LTD: MatchAction Performance testing"

Merge "pkt_gen: MoonGen should use class object for line rate calc"

pkt_gen: MoonGen should use class object for line rate calc

Instead of hard coding line speed numbers for rate calcuation,
MoonGen should use its class object initialized at start time.

JIRA: VSPERF-368

Change-Id: I68c688f8c244df16d98cfd15efabf491a317256b
Signed-off-by: bmichalo <bmichalo@redhat.com>

test_spec: LTD: MatchAction Performance testing

One of the new tests we agreed by working on Etherpad
was MatchAction. Maryam led the development, and many
provided comments there. This is another chance for
comments before we merge.

JIRA: VSPERF-242

Change-Id: I9da74efb703d5f221280567159e4ed6d830eb92c
Signed-off-by: Al Morton <acmorton@att.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Billy O'Mahony<billy.o.mahony@intel.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>

Merge "bugfix: Harmonize test/trial RFC2544 terminology"

hugepages: change default num pages + deallocate

Change the default number of allocated pages to the min 10GB. Deallocate
hugepages if we allocated them on the fly.

Change-Id: If9946cec423dd095c1ca1ab3bee3698e2f025047
Signed-off-by: Maryam Tahhan <maryam.tahhan@intel.com>

hugepage_detect: Change hugepage detection to check for free hugepages

Change hugepage detection to check if hugepages are configured. Also
verifies enough free hugepages are available based on test
configuration. Will result in an exception if not enough hugepages
are found. This performs the check based on the guest memory configs
as well as the dpdk socket configs.

Adds some new methods to check for free hugepages based on all
hugepages or for socket free hugepages.

Adds new logging indicitaing needed and free hugepages as well as info
level event if hugepages are not configured at recommended 1GB size.

JIRA: VSPerf-255

Change-Id: I80d5605308861070c412568d8d40339f01b9718b
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

Merge "namespace_bug: Initial creation of namespace fails"

pkt_gen:  MoonGen incorrectly calculates rx/tx % statistics

For MoonGen, the final results parameters: throughput_rx_percent,
throughput_tx_percent were incorrectly calculated when converting
MoonGen to VSPERF format.  In particular, this would occur when more
than one data stream (single channel unidirectional traffic) was used in
the test. When calculating these percentages, only aggregate throughput
values equaling unidirectional traffic were used (for example, for 64
byte frames bidirectional traffic should use 29.76 Mpps, not 14.88 Mpps)

JIRA: VSPERF-341

Change-Id: Ie62bded7aea023c287697f18c4172b58c320e050
Signed-off-by: bmichalo <bmichalo@redhat.com>

Merge "sriov: PCI IDs must be checked before usage"

Merge "dpdk: Testpmd forwarding mode changes in v16.07"

Merge "install: Missing packages for numa library"

Merge "Xena: Modify aggregate_stats one port no stats when bi-direction"

namespace_bug: Initial creation of namespace fails

Initial creation of namespace fails because var/run/netns does not
exist.

JIRA: VSPERF-366

Change-Id: I02e5aef52e1675f303405b69c7461ca6482903af
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

sriov: PCI IDs must be checked before usage

Order of NIC PCI IDs handling was changed to ensure their
check and possible expansion (by missing domain part)
before their first usage. This avoids a vsperf crash
in case that sriov is configured and short NIC PCI IDs
are used.

JIRA: VSPERF-363

Change-Id: I91f724e9558bce5a92520de72b4bf000f33d2f29
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

dpdk: Testpmd forwarding mode changes in v16.07

Testpmd forwarding mode mac_retry was obsoleted in DPDK v16.07.
Retry option is supported by multiple forwarding modes now.
Former "mac_retry" mode can be replaced by "mac retry" construct.

JIRA: VSPERF-333

Change-Id: I1563427985966264f07d10b1a8351784fe999d74
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

Merge "Hugepages: on the fly allocation."

Merge "pkt_gen: MoonGen start rates not correct"

install: Missing packages for numa library

Recent versions of DPDK (16.xx) depend on numa library.
Thus installation scripts were updated to install both
binary and development packages.

JIRA: VSPERF-360

Change-Id: Idd0eac8758f30a355909bb804e99e650b67bba8b
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

bugfix: Harmonize test/trial RFC2544 terminology

VSPERF framework was updated to follow RFC2544 definition
of 'test' and 'trial' terms. Test parameter 'rfc2544_trials'
was replaced by new parameter 'rfc2544_tests'.
Both code and documentation were updated to reflect
these changes.

JIRA: VSPERF-300

Change-Id: I144a28359fd06394f1f87ef324458bafd5aad45f
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

docs: Update vloop-vnf changelog

Updated version of vloop-vnf image was added into the changelog inside
installation guide.

JIRA: VSPERF-349

Change-Id: I64a1acdf2191858dc282ea6c3b588048b3bdd742
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

Hugepages: on the fly allocation.

If no hugepages are available, try to allocate HUGEPAGE_RAM_ALLOCATION
amount of RAM for hugepages.
JIRA: VSPERF-290

Change-Id: I092b3831254579eeebbe45e85884828f9d749895
Signed-off-by: Mars Toktonaliev <mars.toktonaliev@nokia.com>

Xena: Modify aggregate_stats one port no stats when bi-direction

Xena exception when no stats collected on receive on one port
during bi-directional traffic. Modify aggregate_stats in the
XenaDriver.py, add judge whether the two port both has stats,
if one port has no stats, the result use the port's value which
has stats.

JIRA: VSPERF-327

Change-Id: I18c349d609a2834907f0383dbfbebd5cd281f541
Signed-off-by: liting <tli@redhat.com>

pkt_gen: MoonGen start rates not correct

The VSPERF traffic profile for starting rate is in a
percentage of line rate.  MoonGen programs its
starting rate as Mpps.  Therefore the traffic profile
for VSPERF needs to be converted to a MoonGen format.
There was a hardcoding of 14.88 that was used for the
starting frame rate for all frame sizes.  This needs
to be corrected to take into account all frame sizes.

JIRA: VSPERF-350

Change-Id: Ied71668a560d7d2b03483f9a1192dd2fb4965996
Signed-off-by: bmichalo <bmichalo@redhat.com>

Merge "docs: Enhancement of IxNetwork configuration guide"

docs: Update release notes

Release notes were updated with features introduced by Colorado
release.

JIRA: VSPERF-349

Change-Id: I43179028c973cae4be6746e633c27fd3419e951a
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

Merge "pkt_gen:  Remove commas from MoonGen results"

Merge "xena_deadlock: Fix xena2544.exe deadlock"

pkt_gen:  Remove commas from MoonGen results

Results from other traffic generators are reported as
pure numbers, i.e. 2477356.0000. These values are
consumed and processed by other scripts (e.g. CI or
OPNFV results visualization), so we should keep it
consistent among various traffic generators.

JIRA: VSPERF-342

Change-Id: I305421287eb632707038f2c8bc594fcb255ec939
Signed-off-by: bmichalo <bmichalo@redhat.com>

Merge "pkt_gen: MoonGen incorrectly inserting VLAN tag"

Merge "dpdk: Support of DPDK16.07-rc5 and newer"

docs: Enhancement of IxNetwork configuration guide

Configuration guide has been updated with description of
IxNetwork specific configuration options. New section was added
with description of test results sharing between IxNetwork and
VSPERF.
Additional changes:
* wrong formatting inside installation.rst was fixed
* description of IxNetwork configuration was updated inside ixnet.py

JIRA: VSPERF-302

Change-Id: Ida0e086e69b82fd1f9821d16ea750b2dec1d5aa2
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

Integration docs: HelloWorld and other basic examples.

This patch adds to Integration docs some simple testcase examples
like a HelloWorld. The purpose is to show how to use some basic
features: rules on L3/L4 fields, MultiStream, Multi-VM setup.

Patch Set 2: removed trailing whitespaces.

Patch Set 3: moved the examples to the middle of the file to avoid
potential conflicts with the VxLan TC patch from Sugesh.

Patch Set 4: fix typo.

Patch Set 5: added descriptions on each example.

Patch Set 6: reworked description for VM Replacement TC.

JIRA: VSPERF-331

Change-Id: I9dea23c26138db874fd67f41d8de8a628db5d91f
Signed-off-by: Antonio Fischetti <antonio.fischetti@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Martin Klozik <martinx.klozik@intel.com>

pkt_gen: MoonGen incorrectly inserting VLAN tag

MoonGen needs to take into account the VLAN tag 'enable'
control knob defined within the VSPERF traffic profile
and therefore only insert the VLAN tag when its 'True'.

JIRA: VSPERF-329

Change-Id: Idbeed2f7e6411f0c28aa56a2d3e71bd7a79d975a
Signed-off-by: bmichalo <bmichalo@redhat.com>

dpdk: Support of DPDK16.07-rc5 and newer

DPDK 16.07-rc5 has renamed script tools/dpdk_nic_bind.py
to tools/dpdk-devbind.py. VSPERF was updated to use wildcard
in the script name, which is backward compatible.
Function get_version() from tools/systeminfo was updated
to correctly parse version information for DPDK v16. Version
name includes a minor version as defined inside rte_version.h.
It means, that for tag v16.07-rc5 VSPERF will show v16.07.0-rc5
in the report.

JIRA: VSPERF-333

Change-Id: I61a4ab2d90304fffdfc4d22a5cbf0cb7295402b3
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

xena_deadlock: Fix xena2544.exe deadlock

Fix Xena2544.exe execution with mono where sometimes the
Xena process says complete, but the mono process doesn't exit.

Changes call of Xena2544.exe from communicate thread lock to a
wait try/catch block and checks the log periodically to see
if a deadlock has occured and if so, the mono pid is terminated
to prevent deadlock.

* Fix one small issue with typecasting to int, changed to float

JIRA: VSPERF-319

Change-Id: Ie7a86a2b9ff9cdfacba5a912ad17df6cc0581702
Signed-off-by: Christian Trautman <ctrautma@redhat.com>

docs: Remove trailing whitespaces

Wrong whitespace characters breaks automatic generation of VSPERF
documentation. It is necessary to remove wrong characters before
merge of the draft into VSPERF doc, e.g. by filtering through
tr -d '\014'

JIRA: VSPERF-324

Change-Id: Ie280f5a443c89d4235cae5eb89fdceae74ea4213
Signed-off-by: Martin Klozik <martinx.klozik@intel.com>
Reviewed-by: Maryam Tahhan <maryam.tahhan@intel.com>
Reviewed-by: Al Morton <acmorton@att.com>
Reviewed-by: Christian Trautman <ctrautma@redhat.com>
Reviewed-by: <bmichalo@redhat.com>

Merge "pkt_gen: Moongen class name not consistent with name in conf files"

pkt_gen: Moongen class name not consistent with name in conf files

Need to change naming to 'Moongen' for consistent naming
between conf files and class

JIRA: VSPERF-323

Change-Id: Ic580b68323a759b3c2ef7ca3c07006243a3acf7d
Signed-off-by: bmichalo <bmichalo@redhat.com>