Age | Commit message (Collapse) | Author | Files | Lines |
|
This adds a new API call to add session namespaces
It now takes a netns and a sock_name.
(1) If no netns is passed, sock_name will be used as
socket path. Defaulting to /run/vpp/app_ns_sockets/${ns_id}
(2) If a netns is passed, the sock_name has to be
abstract (i.e. start with '@'). It will default to
`@vpp/session/${ns_id}` and will be created in the provided
netns.
Type: feature
Change-Id: I90e9a8e5ecca2cabe7c05335663e33c8506dc9e7
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
Type: improvement
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
Change-Id: Ic0fa4f83048a280a7d1b04198c0f903798562d2d
|
|
Type: feature
Change-Id: I91f72f5802db195d1a15424d67c1b6e518168f9f
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Type: fix
Currently when a new TAP/TUN device is created from tap_create_if()
via the TUNSETIFF ioctl(), a name is allocated by the kernel (eg.
tap0). If the caller supplied a name this is subsequently set via
netlink, after the device has been created.
Now we request the kernel to create the new device with the caller's
requested name in the first instance, thus avoiding the need to
rename the device, and therefore avoiding a window where the device
exists with a different name.
This can be beneficial, for example, when writing systemd-udevd link
files [1]. Having the TAP/TUN devices created with the requested name
ensures they can be correctly matched by the OriginalName option.
Writing link files might be necessary, for example, to avoid VPP and
systemd-udevd racing to set the MAC address on a newly created TAP
interface. systemd-udevd can be configured to not manipulate the MAC
addresses of matched interfaces.
These changes also resolve an issue where the created device would not
be renamed if the caller requested it be moved to a different network
namespace, since vnet_netlink_set_link_name() was not called in that
case.
[1] https://www.freedesktop.org/software/systemd/man/systemd.link.html
Signed-off-by: Duncan Eastoe <duncan@graphiant.com>
Change-Id: I3d657632856d03979d6b914225c3c841c379e0a1
|
|
Type: fix
The init of fib_index_by_sw_if_index wasn't
setting default value to 0. Which we now
need for setting interfaces unnumbered
Change-Id: Ie5be9b5e5373ef055557a871ad4d1c45fbfc1dee
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Id0e67a9ce9c5d8bca2cd7a30493f03a02a46dc46
|
|
Type: feature
Thought this might be useful when using
many VRF to get the list of allocated
VRFs and corresponding names
Change-Id: If9d2c6612d4215e7576315d66d1eb130fcecfa13
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
Type: fix
When the local MAC address of an interface changes the rewrite strings
of all adjacency types need to be updated - this patch fixes the missing
case of incomplete adjacencies.
I moved the update of all adj types into the adj module, since if the
complete adjs were done by the ip-neighbour module and incomplete ones
by adj module, that would mean two walks of the adj DB, as it is not
possible to walk only a specific type.
UT is updated to include the missing case.
Signed-off-by: Neale Ranns <neale@graphiant.com>
Signed-off-by: Ivan Shvedunov <ivan4th@gmail.com>
Change-Id: I36af94976c645bdd0d4d3bc0093b24d7d077e9d7
|
|
vlib_buffer_enqueue_to_next() requires to allow overflow of up to 63
elements of 'buffer' and 'nexts' array.
- add helper to compute the minimum size
- fix occurences in session and async crypto
Type: fix
Change-Id: If8d7eebc5bf9beba71ba194aec0f79b8eb6d5843
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
vlib_unix_recursive_mkdir () expects a NULL-terminated C-string.
Type: fix
Change-Id: I412b48443d0792307d611c466747c0aa5e423417
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
add CLI "set interface name <current-int-name> <new-int-name>
and the corresponding binary API to allow custom interface name
setting for any interface.
Type: feature
Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: I2b39da59879fd4526bcb5aa5854b6bd21e72ea73
|
|
Change-Id: I5bb1ab3a57097bf03e7de5541cc8496655bc728f
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: fix
Fixes: 5b22339973feaa3eba48ead1e9fb428b1816c0cc
|
|
This change restores the af_packet tx path prior
in use prior to Change
Idf0bdd88990254a614962c2f7bc3e0292ccfd61a but
fixes the stalling issue by ensuring that the next
tx frame pointer is only incremented when a new tx
frame is used. This change also enables the
af_packet PACKET_QDISC_BYPASS option, for better
performance.
Change-Id: I65dde648ed66d21654847a900ceda5a8980ae6ec
Type: improvement
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
|
|
When an input supports adaptive mode and enters
in polling, some rx queues may be missing from
the node's rx_poll_vector. To solve the issue,
a full poll vector is being computed in
vnet_hw_if_update_runtime_data, and returned
by vnet_hw_if_get_rxq_poll_vector when the
node is polling in adaptive mode.
Type: fix
Signed-off-by: Maxime Peim <mpeim@cisco.com>
Change-Id: I249bcb20ae0dd28afb0a5ca32993092bafd2f6b1
|
|
The app may send packet to different remote via sendto() with
same socket (without calling connect()). In such scenario, remote
address need to be filled in.
Type: fix
Signed-off-by: liuyacan <liuyacan@corp.netease.com>
Change-Id: I49deb308668c05442f237e97acc9bfa239782482
|
|
Print header for each table.
Type: improvement
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I54ca04d6000a98afb4ad3d2f24caae3349548f23
|
|
Type: feature
This patch adds IPsec ESP SPI as RSS key for better IPsec
inbound processing scaling. With this feature enabled,
The NIC will use the packets' SPI index as the RSS key
to distribute them to different queues.
The dpdk-input is also updated to support this feature.
Sample CLI command to enable Ipsec ESP SPI as RSS key:
test flow add src-ip any dst-ip any rss types esp
test flow enable eth0 index 0
Signed-off-by: Piotr Bronowski <piotrx.bronowski@intel.com>
Signed-off-by: Gabriel Oginski <gabrielx.oginski@intel.com>
Change-Id: I8b7d01cfc065e9099fad33042ce76898e16ddbf0
|
|
CLIB_PREFETCH (s->tx_fifo, 2 * CLIB_CACHE_LINE_BYTES, LOAD);
sizeof(svm_fifo_t) is 128 bytes
Note on 64B cacheline size Arm machine,
CLIB_CACHE_LINE_BYTES 128
CLIB_CACHE_PREFETCH_BYTES 6
above CLIB_PREFETCH () macro will be expand to
__builtin_prefetch(s->tx_fifo)
__builtin_prefetch(s->tx_fifo + 64)
__builtin_prefetch(s->tx_fifo + 128) << prefetch out of range
__builtin_prefetch(s->tx_fifo + 192) << the same here
Solution:
Change to CLIB_PREFETCH (s->tx_fifo, sizeof (*(s->tx_fifo)), LOAD);
Type: fix
Signed-off-by: Tianyu Li <tianyu.li@arm.com>
Reviewed-by: Lijian Zhang <lijian.zhang@arm.com>
Change-Id: I745cbce3dbe5afcab53c39189d18392f569df5aa
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I54fa6e8f12bb52988888e7019a88d6891017944c
|
|
Tcp may want to send acks in established, but the app
has not called accept() yet.
Type: improvement
Signed-off-by: liuyacan <liuyacan@corp.netease.com>
Change-Id: I43b8cd386e533ca95c8ec260a0a1f695ea140358
|
|
Take advantage of the new TX infra and support manual thread placement
Type: improvement
Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: Id8023846a2eb59125fcf2f80f4f11eb681cf14dc
|
|
Extend the srv6 sid list structure with policy type to support
different SR policy types
Type: feature
Signed-off-by: Ahmed Abdelsalam <ahabdels@cisco.com>
Change-Id: I6a8cf82b1269e1c46f3757df0047b306f613112b
|
|
Type: fix
Signed-off-by: jxm <jiangxiaoming@outlook.com>
Change-Id: I5a6a1ce9f232d4ed0fd4707cd5aa9b46cc8a2a5a
|
|
Type: fix
Change-Id: I9d3344374738f8cf883cffb4715c76b50bb4bfce
Signed-off-by: Stanislav Zaikin <zstaseg@gmail.com>
|
|
- add format_vnet_buffer and format_vnet_buffer_no_chain to mirror
format_vlib_buffer and format_vlib_buffer_no_chain
- format_vnet_buffer used to be the "no chain" version, replace all of
its current use with the corresponding format_vnet_buffer_no_chain
- add a function to dump vnet buffer details from gdb
Type: improvement
Change-Id: I143ce845f80e7ef937ea33a557b6e3b5988c5b8f
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
vhost interrupt mode support with mq does not work when coalesce frame
is not configured to 0.
When packed-ring is configured, we were also setting the wrong flag
for want-interrupt.
No need to trigger an interrupt to RX infra if there is at least one
queue in the same thread that is doing polling.
Type: fix
Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: I8e64250f2aa6bf611a777a53e4d2c1076d83305f
|
|
Remove unused parameters and fix warnings.
Type: fix
Signed-off-by: Klement Sekera <ksekera@cisco.com>
Change-Id: I2d0e7b84b56817999283ecb6be606159dcb26a28
|
|
ip/ip_types_api.h is already installed by the build system
Change-Id: Iec43e0ad50e32473d93ea8f7281b5a6477ddc87f
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: improvement
|
|
Type: fix
Creating a sub-int without exact-match set, and subsequently adding an
IPv4 or IPv6 address will crash VPP. This fix catches this situation and
refuses to allow the caller to add an IPv4 or IPv6 address on an
ethernet sub-int that does not have exact-match set.
TESTED:
Before this change, the following crashes VPP:
```
DBGvpp# cre sub TenGigabitEthernet3/0/0 1 dot1q 10
TenGigabitEthernet3/0/0.1
DBGvpp# set interface ip address TenGigabitEthernet3/0/0.1 2001:db8::1/64
<crash>
```
After the change, VPP refuses to act:
```
DBGvpp# cre sub TenGigabitEthernet3/0/0 1 dot1q 10
TenGigabitEthernet3/0/0.1
DBGvpp# set interface ip address TenGigabitEthernet3/0/0.1 192.0.2.1/30
set interface ip address: sub-interface without exact-match doesn't support IP addressing
DBGvpp# set interface ip address TenGigabitEthernet3/0/0.1 2001:db8:1/64
set interface ip address: sub-interface without exact-match doesn't support IP addressing
```
Signed-off-by: Pim van Pelt <pim@ipng.nl>
Change-Id: I42997db314225cd186ebb54013b5717ace7f7bd6
|
|
Type: feature
Change-Id: Ia970f444ba2f38b7a42ea94942c906f1b541511b
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
Implement a simple source address selection algorithm
for IPv4 and IPv6.
IPv6 does not yet implement RFC6724 but supports link-locals.
ping now chooses correct source address for link-local destination.
Added ping support for link-local multicast (e.g. allnodes).
Type: feature
Signed-off-by: Ole Troan <ot@cisco.com>
Change-Id: I1a3382c1f7d4ace0386c2c19e4e47b045b73a3ed
Signed-off-by: Ole Troan <ot@cisco.com>
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I70791285cbf427479d2dcbf70ffdce2253add2fb
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Idab0f455b52a38efd2cf3781497d8cc7f57bfa42
|
|
than 16-8-8-8
Type: feature
the memory trade-off is:
for 8-8-8-8 tries
DBGvpp# sh ip fib mtrie mem
ipv4-VRF:0 mtrie:8068 hash:3499
totals: mtrie:8068 hash:3499 all:11567
for 16-8-8
DBGvpp# sh ip fib mtrie mem
ipv4-VRF:0 mtrie:333056 hash:3499
totals: mtrie:333056 hash:3499 all:336555
Signed-off-by: Neale Ranns <neale@graphiant.com>
Change-Id: I5271a4322d786de6e47613cff9bd432762dbed2a
|
|
Type: feature
The difference being the MTRIE type they contain.
THE FIB continues to use the 16-8-8 version.
Signed-off-by: Neale Ranns <neale@graphiant.com>
Change-Id: I5a54d4e6e6cc639f18a3fb65ef2925507a7ef1de
|
|
Type: improvement
there's a time-space trade-off between the 16-8-8 and 8-8-8-8 stride.
FIB continues to use the 16-8-8. Other features are now free to make the
choice.
Signed-off-by: Neale Ranns <neale@graphiant.com>
Change-Id: I6691a163486ce62e75e629f6ef0c990f253df8e5
|
|
Type: improvement
the existing mtrie functions use the existing 16_8_8 mtrie. Rename them
to make that explicit. Then we can add the 8_8_8_8 types and functions
alongside.
Signed-off-by: Neale Ranns <neale@graphiant.com>
Change-Id: If5ce7a282d5d3742dc65bcd0523220235d9c510d
|
|
When a packet goes through interface-output, whether it should be
captured or not will be evaluated twice: in interface-output and in
interface-output-template (<interface_name>-out) node.
interface-output should only capture packets if there is an interface
filter, otherwise it should defer evaluation to
interface-output-template.
Type: fix
Change-Id: Ia720c5d184a236335e1d7d44d8c04caf87af5630
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: refactor
as opposed to wrtiing out the mtrie steps one by one each time.
Signed-off-by: Neale Ranns <neale@graphiant.com>
Change-Id: I1248861350f9189f9a67ac6e68940813af279e03
|
|
Type: fix
There's no such thing as a conflict between the FIB and adj tables. This
is code that should have been removed way back in 16.04.
Signed-off-by: Neale Ranns <neale@graphiant.com>
Change-Id: I4524e7921b03b89106dd1f35aa1171d0945367cf
|
|
Even when periodic RAs are disabled VPP would respond to
router solicitations. Making it impossible to have an IPv6
enabled interface with hosts connected to it without VPP
acting as a default router.
This change drops RS messages if the radv_info->send_radv is
off.
Type: fix
Signed-off-by: Ole Troan <ot@cisco.com>
Change-Id: I9a68f8e12c93c1c00125b54f8fd454f48fa22caa
Signed-off-by: Ole Troan <ot@cisco.com>
|
|
l4 match parsing should not try to consume the whole input, otherwise
it breaks cli such as:
'classify session ... match l4 dst_port 22 action set-ip4-fib-id 2'
Type: fix
Change-Id: I81a1b5779811f7df8286a371f85fafe09c947b87
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
- restore fib paths support for ip4
- initialize payload_proto to the relevant default protocol so that
'via <dev>' paths are supported
- fix 'rx all'
- fix temp path vector mem leak
Type: fix
Change-Id: I564d88dc4dce86884ff6791af69974e6d70ff7ca
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Signed-off-by: Filip Tehlar <ftehlar@cisco.com>
Change-Id: I49c4183a443b7b39924328900e6a6ac2e09be426
|
|
The ipsec startup.conf config currently exists in ipsec_tun.c. This is
because currently the only ipsec{...} options are tunnel related.
This patch moves the ipsec config to a common file (ipsec.c) for future
extensibility/addition of non-tunnel related config options.
Type: refactor
Signed-off-by: Zachary Leaf <zachary.leaf@arm.com>
Change-Id: I1569dd7948334fd2cc28523ccc6791a22dea8d32
|
|
forwarding gre payload
This commit introduces 2 fixes:
1) After GRE decapsulation sw_if_index[VLIB_TX] is set as fib index of GRE tunnel.
But since GRE tunnel can work on v4 endpoints and have v6 payload, we need to reset it.
In case we get IPv6 packet inside IPv4 GRE tunnel (or vice-versa) fib index can be (and usually is) invalid.
2) Check that ip-table and ip6-table are the same when setting interface as an unnumbered one.
Also, fix for the pipe test include setting the right unnumbered interface for the pipes
Type: fix
Signed-off-by: Stanislav Zaikin <zstaseg@gmail.com>
Change-Id: Id13d239cfdd21e0db6b1c9725f01c40d4af4d800
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I90c2a191ab34a2a7df3fb0a951e5fc78f40ccfe2
|
|
In high traffic scenarios, if app crashed or hang on somewhere, app_mq
will quickly accumulate to full, after which vpp worker will try 100
times before giving up allocating slot for every msg. This will cause
vpp main thread barrier sync to fail.
Type: fix
Signed-off-by: liuyacan <liuyacan@corp.netease.com>
Change-Id: I2b2bf2b272c5b3ca7e4a56af179af12bbcde149d
|
|
App namespaces can now be associated to a linux ip netns, e.g.:
app ns add id <ns_id> secret <n> sw_if_index <n> netns <netns>
If session layer's app sock api is enabled, this triggers the creation
of an abstract listening socket in the netns that has been configured.
For the example above that would be @vpp/session/<ns_id>.
Consequently, vcl, or other apps attaching to vpp, can connect to said
abstract socket from an ip netns without the need to share unix domain
socket files. In particular, for vcl it's enough to set app-socket-api
to @vpp/session/<ns_id> in the conf file.
Type: feature
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I26fdc626a760a3f423c5b8be4251623f6e9cd73a
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I8e80252b85dda9a8f5699109264dc1b913581442
|