Type: docs
Change-Id: Icfb478427bcbca35162935537ab35ed4f6159dad
Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>

- fixes crash in vpp-debug-verify-master-ubuntu2204-x86_64 CI job
  in test_snort_06_detach_if testcase
- fix similar logic in attach handler
- verify snort direction in attach api message
- add tests verifying attribute validation in attach testcase

Type: fix
Fixes: 1025754

Change-Id: I96fbeb0a7b84f2f238df15b20476ed4086251471
Signed-off-by: Dave Wallace <dwallacelf@gmail.com>
(cherry picked from commit 2b85ab6)

1) we rely on Netlink PSAMPLE and USERSOCK make sure we have headers
2) sflow plugin is not available on FreeBSD, due to Netlink features
3) preprocessor flag SFLOW_USE_VAPI controls whether the sflow plugin
will attempt to contact the linux-cp plugin using the binary VAPI. It
has to be a preprocessor flag so it can suppress the "#include"
statements that define that linux-cp api, because those include files
will be missing if linux-cp is excluded from the build.

The "excluded_plugins" list can be used with:
make VPP_EXCLUDED_PLUGINS=linux-cp build build-release

Type: fix
Fixes: e40f8a9
Change-Id: If44929d285d27db7862910ecb4ec11ddd4591fd2
Signed-off-by: pim@ipng.nl
(cherry picked from commit d9b9a72)

AssertContains and AssertNotContains are now case insensitive to
prevent issues like in HttpPersistentConnectionTest.

Type: test

Change-Id: Ic52e3fa5a7d1d5cb8486e86aa39a845cbd911d94
Signed-off-by: Matus Fabian <matfabia@cisco.com>
(cherry picked from commit d38c132)

Remove the requirement to coordinate between linux-cp and sflow at
build time via cmake. Also, remove 350 lines of awkward thread-forking
VAPI code. Replace it with a dynamically retrieved function from the
linux-cp plugin, called lcp_itf_pair_get_vif_index_by_phy().

Remove build inhibit based on Netlink headers, and only inhibit the
build for FreeBSD. This plugin can now run regardless of Linux CP
being built or loaded, and then return VPP ifIndex numbers instead.

Also, fix a flaky test where non-ephemeral source ports throw off
packet captures.

Type: improvement
Change-Id: I5486742fa0e343e382630a22672a05fe3dcf7836
Signed-off-by: neil.mckee@inmon.com
Signed-off-by: pim@ipng.nl
(cherry picked from commit f1467f6)

Do not set vm.max_map_count to 3096 by default. That value is an order
of magnitude smaller than current distro defaults, and causes
seemingly "random" severe application failures; known to occur when
running Brave, Firefox, and VirtualBox to name but a few.

Type: fix

Change-Id: Ibe7518d92c3be9a4704972c9d504777776f924dc
Signed-off-by: Dave Barach <dave@barachs.net>
(cherry picked from commit 81996c1)

Linux default settings are enough, and modifying them can break other
applications.

Type: improvement

Change-Id: I773b730802bae78b0f33ca3ff31d9ee2ffa82f89
Signed-off-by: Benoît Ganne <bganne@cisco.com>
(cherry picked from commit 5819e2a)

Type: fix

Change-Id: Ic7b9819226742b0f1153ec2794ae80556c49dad2
Signed-off-by: Matus Fabian <matfabia@cisco.com>
(cherry picked from commit 30bd270)

Type: fix

Change-Id: Icabe556448ad7109f4802ad48a701667911652de
Signed-off-by: Adrian Villin <avillin@cisco.com>
(cherry picked from commit d39fa81)

validate segs vector to (_n_segs - 1)

Type: fix

Change-Id: I721cc438e4c9610760a1f40b6355834bbd389b20
Signed-off-by: Matus Fabian <matfabia@cisco.com>
(cherry picked from commit 778de83)

if feature_bitmap is zero, it will make feat_bitmap_get_next_node_index function generate segment fault

Type: fix

Change-Id: I6a9c0b29d927e600537e9e43ad696d09c091f9b1
Signed-off-by: fenglei <1579628578@qq.com>
(cherry picked from commit 03c21e0)

This patch updates the DPDK ICE driver to ensure the correct hardware
engine is selected for flow offload, addressing issues with
improper engine assignment.

Type: fix

Change-Id: Ief7bfbe978d8a684b992ca85b07097112e497b80
Signed-off-by: Kai Ji <kai.ji@intel.com>

Type: fix

Change-Id: I43a221ac2be9bbea885116c12d1991cf4477fae8
Signed-off-by: Matus Fabian <matfabia@cisco.com>
(cherry picked from commit 94f6034)

In CpuAllocator, the default assumption of two node core ranges
seems to not be fully correct. Added handling of multiple ranges
and singular cores.

Type: fix

Change-Id: Id50147c5360baa4035fcd87e3717b0d6c9ea7e5f
Signed-off-by: Semir Sionek <ssionek@cisco.com>
(cherry picked from commit 119ae0a)

Also optimize lookup to avoid session rules table matching.

Type: fix

Change-Id: I5b62c870edd9f7486e7de1417816fffa30d03a3e
Signed-off-by: Florin Coras <fcoras@cisco.com>
(cherry picked from commit 09341c6)

Type: docs
Change-Id: Iff452c28396159f7818e0a294daa0277870077f4
Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>

Adding a virtual MAC to some NICs can take a significant amount of time.
If a lot of VRs enter the master state around the same time, the process
node can stay active for a very long time processing all of the transitions.

Try to force a 10 us sleep between processing events to ensure that the
process node does not prevent API messages and RPCs from being handled for
an extended period.

Type: improvement

Signed-off-by: Matthew Smith <mgsmith@netgate.com>
Change-Id: I400a7d395b4c0fec371f70d359e9d184ae79645e
(cherry picked from commit b41167a)

…ed IPv6 comparison

The comparison code for the bit remainder (non-zero part of prefix length modulo 8)
was incorrectly looking one byte further than it should.

Type: fix
Change-Id: Idd27d218e77eff5f368f2ba0a5cefb86ecf605f5
Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
(cherry picked from commit 7e295a4)