This fixes crash when local socket memory is not available
so rx/tx queue setup fails. Here we simply retry operation
without being so picky about memory location.
Change-Id: I5bae47defe3c0fe120853378ed13141893284a5a
Signed-off-by: Damjan Marion <damarion@cisco.com>
Set max L3 packet size allowed to the lowest value of the slave links.
Also cleanup FIXME's for VMXNET3 driver.
Change-Id: Ia5cf742f2ab13b8cdc8ea6286f56464ee4f8c51d
Signed-off-by: John Lo <loj@cisco.com>
(to be consistent with other delete interface commands)
Change-Id: I41f63cc5b11d9fa9489edc54b17a4d857b711b66
Signed-off-by: Alpesh S. Patel <apatel9191@hotmail.com>
Following two commands are changed:
ethernet mtu -> set interface mtu
ethernet promiscuous -> set inteface promiscuous
Change-Id: I5037e021933156c06044fb723a05ad330f8162b7
Signed-off-by: Damjan Marion <damarion@cisco.com>
Internally change the clearing and displaying of the dpdk stats to be
based on deltas.
Change-Id: I76605ac67492a374ff5522ff44d4a0190cf94e18
Signed-off-by: Todd Foggoa (tfoggoa) <tfoggoa@cisco.com>
gcc version 4.9.2 (Raspbian 4.9.2-10)
Tested on Linux raspberrypi 4.4.6-v7+ #875 SMP Tue Apr 12 16:33:02 BST 2016 armv7l GNU/Linux
CPUs may be little or big endian, detect with gcc flags, not the processor architecture
Add a new flag $(PLATFORM)_uses_openssl which allows to disable the link with openssl lib.
vlib/vlib/threads.c:
startup.conf must:
- specify the heapsize as we don't have hugepages on raspbian
cpu {
main-core 3
}
heapsize 64M
Corrects in various files the assumption uword == u64 and replaces 'u64' cast with 'pointer_to_uword' and 'uword_to_pointer' where appropriate.
256 CPUs may create an OOM when testing with small memory footprint ( heapsize 64M ), allows the number of VLIB_MAX_CPUS to be set in platforms/*.mk
vppinfra/vppinfra/longjmp.S:
ARM - copy r1 (1st parameter of the setjmp call) to r0 (return value)
vppinfra/vppinfra/time.h:
On ARMv7 in AArch32 mode, we can access to a 64bit register to retreive the cycles count.
gcc on rpi only declare ARM_ARCH 6. Override this info, and check if it is possible to use 'mrrc'.
/!\ the time function will NOT work without allowing the user mode access to the PMU.
You may download the source of the kmod here:
https://github.com/christophefontaine/arm_rdtsc
Change-Id: I8142606436d9671a184133b935398427f08a8bd2
Signed-off-by: Christophe Fontaine <christophe.fontaine@qosmos.com>
- introduces default socket-mem size of 512 MB per socket
- default socket-mem value is applied to all discovered CPU sockets
- fixes bug when < 1024 socket-mem parameter is specified
- for socket-mem < 1024 code prefers 2 MB pages
- improves handling of manualy specified socket-mem values
Change-Id: I9ef848108d7dd1228fbd82a5be49eb5277a93683
Signed-off-by: Damjan Marion <damarion@cisco.com>
Need to clear the bitmap entry when processing each input file
descriptor.
This becomes an issue on interface deletion. A crash occurs due
to assertion in a debug image
https://gerrit.fd.io/r/#/c/615/ - removed the code when this was set
again, but the bitmap needs to be cleared
Change-Id: I7cccb9bf8b9b2eb5a7f60ecda7530784a9512116
Signed-off-by: Peter Lei <peterlei@cisco.com>
This is complete rework of DPDK PCI initialization. It drops
previous scheme where lspci/route/awk/sed are used and instead
sysfs is solely used for discovering Ethernet PCI devices. Criteria
for blacklisting device is changed from exsiting routing table entry
to simple interface state obtained by SIOCGIFFLAGS ioctl().
It checks for IFF_UP flag, so as long as interface is declared
up and even when carrier is down interface will be blacklisted.
Change-Id: I59961ddcf1c19c728934e7fe746f343983741bf1
Signed-off-by: Damjan Marion <damarion@cisco.com>
The current mechanism for setting up arp-input and ip6-discover-neighbor
output nodes for interfaces using their interface link up/down callback
function is inefficient and has potential timing issue, as observed for
bonded interface. Now both nodes will setup output interface sw_if_index
in the the sw_if_index[VLIB_TX] field of current packet buffer and then
use the interface-ouput node to tx the packet.
One side effect is that vlib_node_add_next_with_slot() needs to be
modified to allow the same output node-id to be put at the specified
slot, even if another slot contain that same node-id already exist. This
requirement is caused by BVI support where all loopback interfaces set
up as BVIs will have the same output node-id being l2-input while, for
output-interface node, the output slot must match the hw_if_index of the
interface.
Change-Id: I18bd1d4fe9bea047018796f7b8a4d4c20ee31d6e
Signed-off-by: John Lo <loj@cisco.com>
This cleans up vnet folder however it doesn't clean the existing
autogenerated files. I'm not aware of any automated solution to achieve
this without involving git.
Change-Id: Ib29869e08e1d3bd87b7b54bc414f88bd9c632372
Signed-off-by: Florin Coras <fcoras@cisco.com>
VHOST counters are reversed. RX packet count
are shown as TX and vise-versa in "show hardware"
statistics.
Change-Id: Ie1d41f1f0845edf495cdf34ca65466fc307940c6
Signed-off-by: Shesha Sreenivasamurthy <shesha@cisco.com>
This should help when adding new adjacency types
Change-Id: I1832c6b7a80b6bc69ed83423a60511b7932f336f
Signed-off-by: Damjan Marion <damarion@cisco.com>
The mask is increasingly small.
This saves a few cycles and becomes significant when there are many
prefix lengths.
Change-Id: Ibd0c9331f675697bb4e90e8ad617994f83edec9c
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
Unknown hop-by-hop options are currently not processed, which triggers an
infinite loop due to the pointer not advancing further in the header.
Change-Id: Idf9176090e042b17aac1baa25a6cb4beb8c199d8
Signed-off-by: Yoann Desmouceaux <ydesmouc@cisco.com>
PatchSet2: Modify the code according to review comments.
PatchSet3: modify sw_if_index1 in encap.c.
Change-Id: Ic4d3ee19a0ba0fa10568e570a79a3cb85cfbc9ab
Signed-off-by: Hongjun Ni <hongjun.ni@intel.com>
For interfaces which are slave links to a bounded interface, do not
allow sub-interface creation nor interface state to be changed.
Change "show interface" to display interface state as "bond-slave"
for slave links to a bonded interface.
Change "show hardware" to support a "bond" keyword and display slave
links to a bonded interface.
Change-Id: I4db3cae6985bcb1489ab16a07c72c5ee9b2f2dd3
Signed-off-by: John Lo <loj@cisco.com>
Control Plane
-------------
In essence, this introduces basic support for map-request/reply
processing, the logic to generate and consume such messages, including
SMRs, a control-plane backend, consisting of an eid-table, locator and
locator-set tables, and CLI to interact with it. Naturally, we can now
serialize/deserialize LISP specific types: addresses, locators,
mappings, messages. An important caveat is that IPv6 support is not
complete, both for EIDs and RLOCs.
Functionally, the DP forwards all packets it can't handle to the CP
(lisp_cp_lookup node) which takes care of obtaining a mapping for the
packet's destination from a pre-configured map-resolver using the LISP
protocol. The CP then caches this information and programs the DP such
that all new packets with the same destination (or within the covering
prefix) are encapsulated to one of the locators retrieved in the
mapping. Ingress traffic-engineering is not yet supported.
Data Plane
----------
First of all, to enable punting to the CP, when LISP GPE is turned on a
default route that points to lisp_cp_lookup is now inserted. The DP
also exposes an API the CP can use to program forwarding for a given
mapping. This mainly consists in allocating a tunnel and programming the
FIB such that all packets destined to the mapping's prefix are forwarded
to a lisp-gpe encapsulating node.
Another important change done for lisp forwarding is that both source
and destination IP addresses are considered when encapsulating a packet.
To this end, a new FIB/mtrie is introduced as a second stage, src
lookup, post dst lookup. The latter is still done in the IP FIB but for
source-dest entries, in the dest adjacency the lookup_next_index points
to a lisp lookup node and the rewrite_header.sw_if_index points to the
src FIB. This is read by the lisp lookup node which then walks the src
mtrie, finds the associated adjacency, marks the buffer with the index
and forwards the packet to the appropriate next node (typically,
lisp-gpe-encap).
Change-Id: Ibdf52fdc1f89311854621403ccdd66f90e2522fd
Signed-off-by: Florin Coras <fcoras@cisco.com>
Can be used by specifying DPDK_VERSION=16.04-rc2 in the make command line
Change-Id: I657b44d7ca22f1ef57756e7703088020fab12bc6
Signed-off-by: Damjan Marion <damarion@cisco.com>
Fields needed only by specific adj type should
be shared.
Change-Id: I59ee15a29d2f5f527f46910a1a63866b291734c7
Signed-off-by: Damjan Marion <damarion@cisco.com>
