Issue:
After vlan flap, zebra was not marking the selected/best route as installed.
As a result, when a static route was configured with nexthop as directly
connected interface's(vlan) IP, the static route was not being installed
in the kernel since its nexthop was unresolved. The nexthop was marked
unresolved because zebra failed to mark the best route as installed after
interface flap.
This was happening because, in dplane_route_update_internal() if the old and
new context type, and nexthop group id are the same, then zebra doesn't send
down a route replace request to kernel. But, the installed (ROUTE_ENTRY_INSTALLED)
flag is set when zebra receives a response from kernel. Since the
request to kernel was being skipped for the route entry, installed flag
was not being set
Fix:
In dplane_route_update_internal() if the old and new context type, and
nexthop group id are the same, then before returning, installed flag will
be set on the route-entry if it's not set already.
Signed-off-by: Pooja Jagadeesh Doijode <pdoijode@nvidia.com>
During shutdown, the main pthread stops the dplane pthread
before exiting. Don't try to clean up any events scheduled
to the dplane pthread at that point - just let the thread
exit and clean up.
Signed-off-by: Mark Stapp <mjs@labn.net>
Effectively a massive search and replace of
`struct thread` to `struct event`. Using the
term `thread` gives people the thought that
this event system is a pthread when it is not
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
When we are installing the flood entry for a vtep in SVD,
ensure VNI is set on the ctx object so that it gets
sent to the kernel and set appropriately with src_vni.
Signed-off-by: Stephen Worley <sworley@nvidia.com>
dplane_mac_info and dplane_neigh_info is modified to be vni aware.
dplane_rem_mac_add/del dplane_mac_init is modified to be vni aware.
During dplane context update (mac and neigh), we use the vni information
and if set, corresponding netlink attribute NDA_SRC_VNI is set and passed to the
dplane.
Signed-off-by: Sharath Ramamurthy <sramamurthy@nvidia.com>
Locking around the list of providers/plugins is not
helpful - these only change at init time. Clear some SA
warnings by removing the locking.
Signed-off-by: Mark Stapp <mjs@labn.net>
Read from the fpm dplane a route update that will
include status about whether or not the asic was
successfull in offloading the route.
Have this data passed up to zebra for processing and disseminate
this data as appropriate.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
Add the initial step of passing in a dplane context
to reading route netlink messages. This code
will be run in two contexts:
a) The normal pthread for reading netlink messages from
the kernel
b) The dplane_fpm_nl pthread.
The goal of this commit is too just allow a) to work
b) will be filled in in the future. Effectively
everything should still be working as it should
pre this change. We will just possibly allow
the passing of the context around( but not used )
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
In order for a future commit to abstract the dplane_ctx_route_init
so that the kernel can use it, let's move some stuff around
and add a dplane_ctx_route_init_basic that can be used by multiple
different paths
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
create a dplane_ctx_route_init_basic so it can be used
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
Zebra needs the ability to pass this data around.
Add it to the dplanes ability to pass.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
zebra: Add a dplane_ctx_set_flags
The dplane_ctx_set_flags call is missing, we will need it. Add it.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
If we have this semantics:
int ret = FAILURE;
if (foo)
goto done;
....
done:
return ret;
This pattern does us no favors and makes it harder to figure out what is going
on. Let's remove.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
This allows Zebra to manage QDISC, TCLASS, TFILTER in kernel and do cleaning
jobs when it starts up.
Signed-off-by: Siger Yang <siger.yang@outlook.com>
Coverity spotted 3 places where `int ret = XXX` was
being used and FRR was immediately assigning a different
value.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
This commit implements necessary netlink encoders for traffic control
including QDISC, TCLASS and TFILTER, and adds basic dplane operations.
Co-authored-by: Stephen Worley <sworley@nvidia.com>
Signed-off-by: Siger Yang <siger.yang@outlook.com>
When reading a on the fly change of an interested netconf netlink
message. The ifindex and ns_id for the context was being set for the sub structure
but not for the main context data structure and zebra_if_dplane_result
was dropping the result on the floor because it was expecting the ns_id and
the interface id to be in a different spot.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
Multipath route may have mixed nexthops of EVPN and IP unicast. Move
EVPN flag to nexthop to support such cases.
Signed-off-by: Xiao Liang <shaw.leon@gmail.com>
The usage of zebra dplane makes the job asyncronous which implies
that a given job will try to add an iptable, while the second job
will not know that its iptable is the same as the former one.
The below exabgp rules stand for two bgp flowspec rules sent to
the bgp device:
flow {
route {match {
source 185.228.172.73/32;
destination 0.0.0.0/0;
source-port >=49156&<=49159;
}then {redirect 213.242.114.113;}}
route {match {
source 185.228.172.73/32;
destination 0.0.0.0/0;
source-port >=49160&<=49163;
}then {redirect 213.242.114.113;}}
}
This rule creates a single iptable, but in fact, the same iptable
name is appended twice. This results in duplicated entries in the
iptables context. This also results in contexts not flushed, when
BGP session or 'flush' operation is performed.
iptables-save:
[..]
-A PREROUTING -m set --match-set match0x55baf4c25cb0 src,src -g match0x55baf4c25cb0
-A PREROUTING -m set --match-set match0x55baf4c25cb0 src,src -g match0x55baf4c25cb0
-A match0x55baf4c25cb0 -j MARK --set-xmark 0x100/0xffffffff
-A match0x55baf4c25cb0 -j ACCEPT
-A match0x55baf4c25cb0 -j MARK --set-xmark 0x100/0xffffffff
-A match0x55baf4c25cb0 -j ACCEPT
[..]
This commit addresses this issue, by checking that an iptable
context is not already being processed. A flag is added in the
original iptable context, and a check is done if the iptable
context is not already being processed for install or uinstall.
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
Avoid initialization in dplane_ctx_intf_init() so
the compiler can warn us about using unintialized data.
Signed-off-by: Stephen Worley <sworley@nvidia.com>
Only clear protodown reason on shutdown/sweep, retain protodown
state.
This is to retain traditional and expected behavior with daemons
like vrrpd setting protodown. They expet it to be set on shutdown
and retained on bring up to prevent traffic from being dropped.
We must cleanup our reason code though to prevent us from blocking
others.
Signed-off-by: Stephen Worley <sworley@nvidia.com>
Add enums for set/unset of prodown state to handle the mainthread
knowing an update is already queued without actually marking it
as complete.
This is to make the logic confirm a bit more with other parts of the code
where we queue dplane updates and not update our internal structs until
success callback is received.
Signed-off-by: Stephen Worley <sworley@nvidia.com>
Add support for setting the protodown reason code.
829eb208e8
These patches handle all our netlink code for setting the reason.
For protodown reason we only set `frr` as the reason externally
but internally we have more descriptive reasoning available via
`show interface IFNAME`. The kernel only provides a bitwidth of 32
that all userspace programs have to share so this makes the most sense.
Since this is new functionality, it needs to be added to the dplane
pthread instead. So these patches, also move the protodown setting we
were doing before into the dplane pthread. For this, we abstract it a
bit more to make it a general interface LINK update dplane API. This
API can be expanded to support gernal link creation/updating when/if
someone ever adds that code.
We also move a more common entrypoint for evpn-mh and from zapi clients
like vrrpd. They both call common code now to set our internal flags
for protodown and protodown reason.
Also add debugging code for dumping netlink packets with
protodown/protodown_reason.
Signed-off-by: Stephen Worley <sworley@nvidia.com>
Use the dataplane to query and read interface NETCONF data;
add netconf-oriented data to the dplane context object, and
add accessors for it. Add handler for incoming update
processing.
Signed-off-by: Mark Stapp <mstapp@nvidia.com>
Currently when the kernel sends netlink messages to FRR
the buffers to receive this data is of fixed length.
The kernel, with certain configurations, will send
netlink messages that are larger than this fixed length.
This leads to situations where, on startup, zebra gets
really confused about the state of the kernel. Effectively
the current algorithm is this:
read up to buffer in size
while (data to parse)
get netlink message header, look at size
parse if you can
The problem is that there is a 32k buffer we read.
We get the first message that is say 1k in size,
subtract that 1k to 31k left to parse. We then
get the next header and notice that the length
of the message is 33k. Which is obviously larger
than what we read in. FRR has no recover mechanism
nor is there a way to know, a priori, what the maximum
size the kernel will send us.
Modify FRR to look at the kernel message and see if the
buffer is large enough, if not, make it large enough to
read in the message.
This code has to be per netlink socket because of the usage
of pthreads. So add to `struct nlsock` the buffer and current
buffer length. Growing it as necessary.
Fixes: #10404
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
Store the fd that corresponds to the appropriate `struct nlsock` and pass
that around in the dplane context instead of the pointer to the nlsock.
Modify the kernel_netlink.c code to store in a hash the `struct nlsock`
with the socket fd as the key.
Why do this? The dataplane context is used to pass around the `struct nlsock`
but the zebra code has a bug where the received buffer for kernel netlink
messages from the kernel is not big enough. So we need to dynamically
grow the receive buffer per socket, instead of having a non-dynamic buffer
that we read into. By passing around the fd we can look up the `struct nlsock`
that will soon have the associated buffer and not have to worry about `const`
issues that will arise.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
