hash: slice CRC data into 8-byte pieces

Calculating hash for data of variable length is more efficient
when that data is sliced into 8-byte pieces. The rest part of data
is hashed using CRC32 functions with either 8 and 4 byte operands.

Signed-off-by: Yerden Zhumabekov <e_zhumabekov@sts.kz>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

hash: fallback to software CRC32 implementation

Initially, SSE4.2 support is detected via the constructor function.

Added rte_hash_crc_set_alg() function to detect and set CRC32
implementation if necessary. SSE4.2 is allowed by default.

rte_hash_crc_*byte() functions reworked so they choose available
CRC32 implementation in the runtime.

Signed-off-by: Yerden Zhumabekov <e_zhumabekov@sts.kz>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

hash: add CRC function for 8 bytes

SSE4.2 provides CRC32 intrinsic with 8-byte operand.

Signed-off-by: Yerden Zhumabekov <e_zhumabekov@sts.kz>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

hash: replace built-in functions implementing SSE4.2

Give up using built-in intrinsics and use our own assembly
implementation. Remove #include entry as well.

Signed-off-by: Yerden Zhumabekov <e_zhumabekov@sts.kz>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

hash: add assembly implementation of CRC32 intrinsics

Added:
- crc32c_sse42_u32() emits 'crc32l' asm instruction;
- crc32c_sse42_u64() emits 'crc32q' asm instruction;
- crc32c_sse42_u64_mimic(), wrapper in case of run on 32-bit platform.

Signed-off-by: Yerden Zhumabekov <e_zhumabekov@sts.kz>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

hash: add software CRC32 implementation

Add lookup tables for CRC32 algorithm, crc32c_1word() and
crc32c_2words() functions returning hash of 32-bit and 64-bit
operand.

Signed-off-by: Yerden Zhumabekov <e_zhumabekov@sts.kz>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

app/testpmd: support NVGRE in Tx checksum offload

Enhance csum fwd engine based on current TX checksum framework in order
to test TX Checksum offload for NVGRE packet.

It includes:
 - IPv4 and IPv6 packet
 - outer L3, inner L3 and L4 checksum offload for Tx side.

Signed-off-by: Jijiang Liu <jijiang.liu@intel.com>
Signed-off-by: Declan Doherty <declan.doherty@intel.com>

app/testpmd: support NVGRE in Rx tunnel filtering

Extend the "tunnel_filter" command in testpmd to test the RX tunnel filter API for NVGRE packet.

Signed-off-by: Jijiang Liu <jijiang.liu@intel.com>
Signed-off-by: Declan Doherty <declan.doherty@intel.com>

i40e: support NVGRE in Rx tunnel filtering

The filter types supported are listed below for NVGRE packet:
   1. Inner MAC and Inner VLAN ID.
   2. Inner MAC address, inner VLAN ID and tenant ID.
   3. Inner MAC and tenant ID.
   4. Inner MAC address.
   5. Outer MAC address, tenant ID and inner MAC address.

Signed-off-by: Jijiang Liu <jijiang.liu@intel.com>
Signed-off-by: Declan Doherty <declan.doherty@intel.com>

ether: add transparent ethernet bridging type

Add an Ethernet type definition for Transparent Ethernet Bridging.

Signed-off-by: Jijiang Liu <jijiang.liu@intel.com>
Signed-off-by: Declan Doherty <declan.doherty@intel.com>

app/testpmd: support new rss offloads

RSS offloads supported 'ip' and 'udp' only, which did not demonstrate
all of the hardware capabilities. The modifications adds support of
new RSS offloads of 'tcp', 'sctp', 'ether' and 'all'.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

app/testpmd: fix some indent

Added code style fixes.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

ethdev: unification of RSS offload types

RSS offload types were defined separately for 1/10G and 40G NICs,
and have no relationship with flow types. The modifications are to
unify all RSS offload types for all PMDs. Unified RSS offload types
have new and common names which can be used for any PMD or
applications, and decouple from specific hardwares.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>
[Thomas: merge with fm10k]
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>

ethdev: unification of flow types

Flow types was defined actually for i40e hardware specifically,
and wasn't able to be used for defining RSS offload types of all
PMDs. It removed the enum flow types, and uses macros instead
with new names. The new macros can be used for defining RSS
offload types later. Also modifications are made in i40e and
testpmd accordingly.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>
[Thomas: merge with new flow director API]
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>

ethdev: fix size of flow type mask array

It wrongly calculates the size of the flow type mask array. The fix
is to align the flow type maximum index ID with the number of
element bit width, and then divide the number of element bit width.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

ethdev: minor comment changes

Added code style fixes.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

i40e: remove some useless line breaks

Added code style fixes.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

ethdev: remove old ethertype filter ABI

The old ethertype filter API was removed in commit 75db20648,
but was still in (newly integrated) version map for ABI.

Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>

ethdev: remove old ntuple filter API

Following structures are removed:
 - rte_2tuple_filter
 - rte_5tuple_filter
Following APIs are removed:
 - rte_eth_dev_add_2tuple_filter
 - rte_eth_dev_remove_2tuple_filter
 - rte_eth_dev_get_2tuple_filter
 - rte_eth_dev_add_5tuple_filter
 - rte_eth_dev_remove_5tuple_filter
 - rte_eth_dev_get_5tuple_filter
It also move macros TCP_*_FLAG to rte_eth_ctrl.h, and removes the macro
TCP_UGR_FLAG which is duplicated with TCP_URG_FLAG.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
[Thomas: remove also from version map]

app/testpmd: new commands for ntuple filter

Following commands of 5tuple and 2tuple filter are removed:
 - add_2tuple_filter (port_id) protocol (pro_value) (pro_mask)
   dst_port (port_value) (port_mask) flags (flg_value) priority (prio_value)
   queue (queue_id) index (idx)
 - remove_2tuple_filter (port_id) index (idx)
 - get_2tuple_filter (port_id) index (idx)
 - add_5tuple_filter (port_id) dst_ip (dst_address) src_ip (src_address)
   dst_port (dst_port_value) src_port (src_port_value) protocol (protocol_value)
   mask (mask_value) flags (flags_value) priority (prio_value)"
   queue (queue_id) index (idx)
 - remove_5tuple_filter (port_id) index (idx)
 - get_5tuple_filter (port_id) index (idx)

New commands are added for 5tuple and 2tuple filter by using filter_ctrl API
and new ntuple filter structure:
 - 2tuple_filter (port_id) (add|del)
   dst_port (dst_port_value) protocol (protocol_value)
   mask (mask_value) tcp_flags (tcp_flags_value)
   priority (prio_value) queue (queue_id)
 - 5tuple_filter (port_id) (add|del)
   dst_ip (dst_address) src_ip (src_address)
   dst_port (dst_port_value) src_port (src_port_value)
   protocol (protocol_value)
   mask (mask_value) tcp_flags (tcp_flags_value)
   priority (prio_value) queue (queue_id)

Test report: http://dpdk.org/ml/archives/dev/2015-February/013049.html

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Tested-by: Huilong Xu <huilongx.xu@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

ixgbe: migrate ntuple filter to new API

This patch defines new functions dealing with ntuple filters which is
corresponding to 5tuple in HW.
It removes old functions which deal with 5tuple filters.
Ntuple filter is dealt with through entrance ixgbe_dev_filter_ctrl.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

igb: migrate ntuple filter to new API

This patch defines new functions dealing with ntuple filters which is
corresponding to 2tuple filter for 82580 and i350 in HW, and to 5tuple
filter for 82576 in HW.
It removes old functions which deal with 2tuple and 5tuple filters in igb driver.
Ntuple filter is dealt with through entrance eth_igb_filter_ctrl.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

ethdev: new ntuple filter API

This patch defines ntuple filter type RTE_ETH_FILTER_NTUPLE and its structure rte_eth_ntuple_filter.
It also corrects the typo TCP_UGR_FLAG to TCP_URG_FLAG

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

ethdev: remove old syn filter API

Structure rte_syn_filter is removed.
Following APIs are removed:
  - rte_eth_dev_add_syn_filter
  - rte_eth_dev_remove_syn_filter
  - rte_eth_dev_get_syn_filter

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
[Thomas: remove also from version map]

app/testpmd: new commands for syn filter

Following commands of syn filter are removed:
  - add_syn_filter (port_id) priority (high|low) queue (queue_id)
  - remove_syn_filter (port_id)
  - get_syn_filter (port_id)
New command is added for syn filter by using filter_ctrl API and new
syn filter structure:
  - syn_filter (port_id) (add|del) priority (high|low) queue (queue_id)

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>

ixgbe: migrate syn filter to new API

This patch defines new functions dealing with syn filter.
It removes old functions which deal with syn filter.
Syn filter is dealt with through entrance ixgbe_dev_filter_ctrl.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>

igb: migrate syn filter to new API

This patch defines new functions dealing with syn filter.
It removes old functions of syn filter in igb driver.
Syn filter is dealt with through entrance eth_igb_filter_ctrl.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>

ethdev: new syn filter API

This patch defines syn filter type RTE_ETH_FILTER_SYN and its structure rte_eth_syn_filter.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>

ethdev: remove old flex filter API

Structure rte_flex_filter is removed.
Following APIs are removed:
  - rte_eth_dev_add_flex_filter
  - rte_eth_dev_remove_flex_filter
  - rte_eth_dev_get_flex_filter

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
[Thomas: remove also from version map]

app/testpmd: new commands for flex filter

Following commands of flex filter are removed:
  - add_flex_filter (port_id) len (len_value) bytes (bytes_string) mask (mask_value)
    priority (prio_value) queue (queue_id)
  - remove_flex_filter (port_id) index (idx)
  - get_flex_filter (port_id) index (idx)
New command is added for flex filter by using filter_ctrl API and new flex filter structure:
  - flex_filter (port_id) (add|del) len (len_value) bytes (bytes_value) mask (mask_value)
    priority (prio_value) queue (queue_id)

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

igb: migrate flex filter to new API

This patch defines new functions dealing with flex filter.
It removes old functions of flex filter in igb driver.
Syn filter is dealt with through entrance eth_igb_filter_ctrl.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>

ethdev: new flex filter API

This patch defines flex filter type RTE_ETH_FILTER_FLEXIBLE and its structure rte_eth_flex_filter.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>

app/testpmd: set default flow director mask

This patch sets the default value of flow director's mask.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

app/testpmd: update display of flow director information

update the function to print information includes:
 - capability
 - mask
 - flex configuration

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

app/testpmd: update flow director commands

Add new command to set flow director's mask:
  - flow_director_mask
Update arguments of commands:
  - flow_director_filter
  - flow_director_flex_mask
  - flow_director_flex_payload
Following commands of flow director filter are removed:
  - add_signature_filter
  - upd_signature_filter
  - rm_signature_filter
  - add_perfect_filter
  - upd_perfect_filter
  - rm_perfect_filter
  - set_masks_filter
  - set_ipv6_masks_filter

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ixgbe: support flow director flush

This patch implement RTE_ETH_FILTER_FLUSH operation to delete all
flow director filters in ixgbe driver.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ixgbe: migrate flow director info and statistic to new API

This patch changes the get info operation to be implemented through
filter_ctrl API and RTE_ETH_FILTER_INFO/RTE_ETH_FILTER_STATS ops.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ixgbe: support new flow director masks

This patch implement the mask configuration of flow director filter,
by using the mask defined in rte_fdir_conf instead of callback function
fdir_set_masks.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ethdev: add new flow director masks

This patch defines structure rte_eth_fdir_masks.
It extends rte_fdir_conf and rte_eth_fdir_info to contain mask's configuration.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ethdev: remove flexbytes offset from flow director

This patch removes the flexbytes_offset from rte_fdir_conf, because
the flexible payload setting is done by flex_conf instead of flexbytes_offset.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ixgbe: support flexpayload configuration of flow director

This patch implement the flexpayload configuration of flow director filter.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ethdev: extend flow type and flexible payload type for flow director

This patch adds RTE_ETH_FLOW_TYPE_RAW and RTE_ETH_RAW_PAYLOAD to support the
flexible payload is started from the beginning of the packet.

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ixgbe: migrate flow director filtering to new API

This patch changes the add/delete/update operations to be implemented through
filter_ctrl API and RTE_ETH_FILTER_ADD/RTE_ETH_FILTER_DELETE/RTE_ETH_FILTER_UPDATE ops.
It also removes the callback functions:
 - ixgbe_eth_dev_ops.fdir_add_signature_filter
 - ixgbe_eth_dev_ops.fdir_update_signature_filter
 - ixgbe_eth_dev_ops.fdir_remove_signature_filter
 - ixgbe_eth_dev_ops.fdir_add_perfect_filter
 - ixgbe_eth_dev_ops.fdir_update_perfect_filter
 - ixgbe_eth_dev_ops.fdir_remove_perfect_filter

Signed-off-by: Jingjing Wu <jingjing.wu@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

tools: enable binding device to uio_pci_generic

Add uio_pci_generic to the list of supported kernel drivers.

Signed-off-by: Danny Zhou <danny.zhou@intel.com>
Tested-by: Qun Wan <qun.wan@intel.com>
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

eal/linux: toggle interrupt for uio_pci_generic

enable/disable interrupt by manipulating a control bit of command
register on NIC's PCIe configuration space.

Signed-off-by: Danny Zhou <danny.zhou@intel.com>
Tested-by: Qun Wan <qun.wan@intel.com>
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

eal/linux: enable uio_pci_generic support

Change the EAL PCI code so that it can work with both the
uio_pci_generic in-tree driver, as well as the igb_uio
DPDK-specific driver.

This involves changes to
1) Modify method of retrieving BAR resource mapping information
2) Mapping using resource files in /sys rather than /dev/uio*
2) Setup bus master bit in NIC's PCIe configuration space for
uio_pci_generic.

Signed-off-by: Danny Zhou <danny.zhou@intel.com>
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

app/test: add unit tests for link bonding mode 6

Added 4 unit tests checking link bonding mode 6 behavior.

Also modified virtual_pmd so it is possible to provide packets,
that should be received with rx_burst and to inspect packets
transmitted by tx_burst.

In packet_burst_generator.c function creating eth_header is
modified, so it accepts ether_type as a parameter and function
creating arp_header is added. Updated other unit tests to get
rid of compilation errors.

Signed-off-by: Maciej Gajdzica <maciejx.t.gajdzica@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

bond: rename mode 5

This patch modify mode older name from
BONDING_MODE_ADAPTIVE_TRANSMIT_LOAD_BALANCING to BONDING_MODE_TLB
This patch also changes order of TEST_ASSERT macro in
test_tlb_verify_slave_link_status_change_failover.

Signed-off-by: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

examples/bond: add example application for link bonding mode 6

This patch contains an example for link bonding mode 6.
It interact with user by a command prompt. Available commands are:
Start - starts ARP_thread which respond to ARP_requests and sends
ARP_updates (this
Is enabled by default after startup),
Stop  -stops ARP_thread,
Send count ip - send count ARP requests for IP,
Show - prints basic bond information, like IPv4 statistics from clients
Help,
Quit.
The best way to test mode 6 is to use this example together with
previous patch:
[PATCH 3/4] bond: add debug info for mode 6 link bonding.
Connect clients thru switch to bonding machine and send:
arping -c 1 bond_ip or
generate IPv4 traffic to bond_ip (IPv4 traffic from different clients
should be then balanced on slaves in round robin manner).

Signed-off-by: Michal Jastrzebski <michalx.k.jastrzebski@intel.com>
Signed-off-by: Maciej Gajdzica <maciejx.t.gajdzica@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

bond: add debug info for mode 6

This patch add some debug information when using link bonding mode 6.
It prints basic information about ARP packets on RX and TX (MAC, ip,
packet number, arp packet type).
If CONFIG_RTE_LIBRTE_BOND_DEBUG_ALB == y.
If CONFIG_RTE_LIBRTE_BOND_DEBUG_ALB_L1 is enabled instead of previous
one, use show command to see IPv4 balancing from clients.

Signed-off-by: Michal Jastrzebski <michalx.k.jastrzebski@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

bond: add mode 6

This mode includes adaptive TLB and receive load balancing (RLB). In RLB
the bonding driver intercepts ARP replies send by local system and
overwrites its source MAC address, so that different peers send data to
the server on different slave interfaces. When local system sends ARP
request, it saves IP information from it. When ARP reply from that peer
is received, its MAC is stored, one of slave MACs assigned and ARP reply
send to that peer.

Signed-off-by: Maciej Gajdzica <maciejx.t.gajdzica@intel.com>
Signed-off-by: Michal Jastrzebski <michalx.k.jastrzebski@intel.com>
Signed-off-by: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

net: change arp header struct declaration

Changed MAC address type from uint8_t[6] to struct ether_addr and IP
address type from uint8_t[4] to uint32_t to make it consistent with
other DPDK code using MAC and IP addresses. It allows us to use
is_same_ether_addr and ether_addr_copy functions on MAC addresses in ARP header.  Also
removed union from arp_hdr struct to make calls to arp_data items
shorter. Updated test-pmd to match new arp_hdr version.

Signed-off-by: Maciej Gajdzica <maciejx.t.gajdzica@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>
[Thomas: doxygenize comments]

virtio: remove Rx hotspots from early return

Remove those hotspots which is unnecessary when early returning occurs.

Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: use port IO to get PCI resource

Make virtio not require UIO for some security reasons, this is to match
6WIND's virtio-net-pmd.

Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: free mbuf's with threshold

This makes virtio driver work like ixgbe. Transmit buffers are
held until a transmit threshold is reached. The previous behavior
was to hold mbuf's until the ring entry was reused which caused
more memory usage than needed.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: set MAC address

Need to have do special things to set default mac address.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: suport multiple MAC addresses

Virtio support multiple MAC addresses.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: support vlan filtering

Virtio supports vlan filtering.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: fix states handling during initialization

Change order of initialization to match Linux kernel.
Don't blow away control queue by doing reset when stopped.

Calling dev_stop then dev_start would not work.
Dev_stop was calling virtio reset and that would clear all queues
and clear all feature negotiation.
Resolved by only doing reset on device removal.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: move allocation before initialization

If allocation fails, don't want to leave virtio device stuck
in middle of initialization sequence.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

examples/vhost: add vlan strip command line option

Support turn on/off RX VLAN strip on host, this let guest get the chance of
using its software VLAN strip functionality.

Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

examples/vhost: avoid inserting vlan twice

Check if it has already been vlan-tagged packet, if true, avoid inserting a
duplicated vlan tag into it.

This is a possible case when guest has the capability of inserting vlan tag.

Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: use soft vlan strip with mergeable Rx packets

To keep the consistent logic with normal Rx path, the mergeable
Rx path also needs software vlan strip/decap if it is enabled.

Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: use soft vlan encap/decap

Implement VLAN stripping in software. This allows application
to be device independent.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

ether: add soft vlan encap/decap

It is helpful to allow device drivers that don't support hardware
VLAN stripping to emulate this in software. This allows application
to be device independent.

Avoid discarding shared mbufs. Make a copy in rte_vlan_insert() of any
packet to be tagged that has a reference count > 1.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: remove redundant alignment field

Since vq_alignment is constant (always 4K), it does not
need to be part of the vring struct.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: remove unnecessary adapter structure

Cleanup virtio code by eliminating unnecessary nesting of
virtio hardware structure inside adapter structure.
Also allows removing unneeded macro, making code clearer.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: support link state interrupt

Virtio has link state interrupt which can be used.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: allow starting with link down

Starting driver with link down should be ok, it is with every
other driver. So just allow it.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: fix update of vring descriptor index

Updating the vring descriptor index should be done before notifying host;
Remove 2 duplicated store memory barriers in both Rx and Tx path because there is
store memory barrier in vq_update_avail_idx function;
Notify the host only if packets actually transmitted(nb_tx > 0).

Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: use weaker barriers

The DPDK driver only has to deal with the case of running on PCI
and with SMP. In this case, the code can use the weaker barriers
instead of using hard (fence) barriers. This will help performance.
The rationale is explained in Linux kernel virtio_ring.h.

To make it clearer that this is a virtio thing and not some generic
barrier, prefix the barrier calls with virtio_.

Add missing (and needed) barrier between updating ring data
structure and notifying host.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: check packet headroom at compile time

Better to check at compile time than fail at runtime.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: make a function local

Make vtpci_get_status a local function as it is used in one file.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

virtio: rearrange resource initialization

For clarity make the setup of PCI resources for Linux into a function rather
than block of code #ifdef'd in middle of dev_init.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Changchun Ouyang <changchun.ouyang@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

i40e: fix build with gcc 5

Eliminate ambiguity in the condition which trips up a "logical not
is only applied to the left..." warning from gcc 5, causing build
failure with -Werror. Besides non-ambiguous, the condition is
far more obvious this way.

Signed-off-by: Panu Matilainen <pmatilai@redhat.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

app/testpmd: fix uninitialized flow control variables

rx_fc_en and tx_fc_en in cmd_link_flow_ctrl_set_parsed
could be used without being initialized.

Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

ixgbe: forbid building vpmd without Rx bulk alloc

CONFIG_RTE_LIBRTE_IXGBE_RX_ALLOW_BULK_ALLOC is a prerequisite
of CONFIG_RTE_IXGBE_INC_VECTOR.

Reported-by: Alexander Belyakov <abelyako@gmail.com>
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>

ixgbe: fix vector receive of chained mbuf

When the vector pmd was receiving a mix of packets of various sizes,
some of which were split across multiple mbufs, there was an issue
with reassembly of the jumbo frames. This was due to a skipped increment
when using "continue" in a while loop. Changing the loop to a "for"
loop fixes this problem, by ensuring the increment is always performed.

Reported-by: Prashant Upadhyaya <praupadhyaya@gmail.com>
Reported-by: Martin Weiser <martin.weiser@allegro-packets.com>
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Tested-by: Martin Weiser <martin.weiser@allegro-packets.com>

ixgbe: fix big endian access

ixgbe is little endian, but cpu maybe not.
add necessary conversions.
    rte_cpu_to_le_32(...) for PCI write
    rte_le_to_cpu_32(...) for PCI read.

Signed-off-by: Xuelin Shi <xuelin.shi@freescale.com>
Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>

e1000: fix big endian access

e1000 is little endian, but cpu maybe not.
add necessary conversions.

rte_cpu_to_le_32(...) for PCI write
rte_le_to_cpu_32(...) for PCI read.

Signed-off-by: Xuelin Shi <xuelin.shi@freescale.com>
Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>

eal/linux: fix fscanf format mismatch

Variables are unsigned int but format scans for signed int.

Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

eal: fix dynamic link with virtio

Building shared libraries and using virtio PMD results in undefined
reference to 'rte_eal_iopl_init'.

Add missing function to eal version map.

Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

cmdline: fix check in port list parsing

The argument ressize contains the size of the result buffer which
should be large enough to store the parsed result of a token. In
this case, it should be larger or equal to sizeof(cmdline_portlist_t)
(4 bytes), not PORTLIST_TOKEN_SIZE which is the max size of the token
string.

This is not a critical, it fixes cases where the total length of the
parsed instruction is greater than the maximum.

Signed-off-by: Olivier Matz <olivier.matz@6wind.com>

doc: add requirements for x32 ABI

Add requirements about compiler and distribution support.

Signed-off-by: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

mk: support x32 ABI

x32 ABI provides benefits of x86-64 while using 32-bit pointers and
avoiding overhead of 64-bit pointers.

Test report: http://dpdk.org/ml/archives/dev/2015-February/012599.html

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Signed-off-by: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
Tested-by: Haifeng Tang <haifengx.tang@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

bond: change warning

Remove function name from warning.

Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

bond: fix symbol export

rte_eth_bond_8023ad_setup and rte_eth_bond_8023ad_conf_get entries need
to be exported to be used by test application for shared libraries
compilation.

Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

app/test: link with ring pmd when needed

This patch links test application against librte_pmd_ring.so for shared
libraries. It's required as long as librte_pmd_ring provides some additional
routines used for testing purposes and must be "hard-linked".

Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

app/test: unit tests for bonding mode 4

This patch adds unit tests for mode 4. It is split into separate
file to avoid problems with other modes that does not need to
look into packets payload.
This patch includes also a modification of maximum number of ports
used in their tests for bonding modes 0-3 from 16 to 6.

Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

app/test: rework assert macros

Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

ring: fix MAC per device

Initialization procedure fix to allow per device MAC configuration.

Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

ring: add MAC address add/remove

Added MAC addr add/remove dummy callbacks.

Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

ring: add link up/down

Link up and down implementation.

Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

bond: fix memory leak on kvargs processing failure

identified by klockwork scan

Signed-off-by: Declan Doherty <declan.doherty@intel.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

ethdev: return status of stats read

rte_eth_stats_get is to get physical device statistics. Without
return status, caller does not know whether function fails or not
(i.e. invalid port_id, driver has no stats_get callback). Caller
cannot differiente normal 0 stats or error case. This fix adds
a return status to the function.

Signed-off-by: Jia Yu <jyu@vmware.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

ethdev: update link status after start

Since LSR interrupt is disabled by pmd drivers, link status in rte_eth_device is always down.
Bond slave_configure() enables LSR interrupt on devices to get notification if link status
changes. However, the LSC interrupt at device start time is still lost.

In this fix, call link_update to read link status from hardware
register at device start time.

Signed-off-by: Jia Yu <jyu@vmware.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

maintainers: add and claim reorder

Add files related to reorder library and claim it.

Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>

doc: new packet ordering app description

This patch describes how to build and run he new packet ordering sample
application that exercises the reorder library.

Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

doc: new reorder library description

This patch introduces a new section in the programmers guide describing
the reorder library.

Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

examples/packet_ordering: new sample app

This new app makes use of the librte_reorder library.

It requires at least 3 lcores for RX, Workers (1 or more) and TX threads.
Communication between RX-Workers and Workers-TX is done by using rings.
The flow of mbufs is the following:
 * RX thread gets mbufs from driver, set sequence number and enqueue
   them in ring.
 * Workers dequeue mbufs from ring, do some 'work' and enqueue mbufs in
   ring.
 * TX dequeue mbufs from ring, inserts them in reorder buffer, drains
   mbufs from reorder and sends them to the driver.

Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>
[Thomas: add in examples/Makefile]