Age | Commit message (Collapse) | Author | Files | Lines |
|
VPP IPv6 FIB relies on 2 bihash for all its VRFs: 1 for control plane
and one for data plane. Because of that, control plane VRF walks
time grows with the number of VRFs as it has to go through all the VRFs.
In order to improve control plane performance with large number of VRFs,
the control plane data structure is changed from a global bihash to a
per-VRF hashtable, similarly to IPv4. Contrary to IPv4, there is only 1
hashtable per VRF and not 1 hashtable per prefix-length per VRF, which
is a compromise between memory consumption and single VRF scaling.
Type: improvement
Change-Id: Ib738a44adb88e46c95c4c62c853ee1f1392c7187
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Destroying a non-empty fib table can lead to difficult to diagnose bugs,
like dandling fib entries preventing entries to be inserting to the
mtrie.
Always log an error to help diagnose those issues.
Type: improvement
Change-Id: I2c771a80595035b440931b74cca4429af22161a2
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: I8c248d9e224bd069b641a174da57d448371470af
Signed-off-by: Jon Loeliger <jdl@netgate.com>
|
|
Type: refactor
Change-Id: I5235bf3e9aff58af6ba2c14e8c6529c4fc9ec86c
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
ip6 and mpls fib_table->ft_locks memory leaked when the table is
deleted.
name tag is leaked for mpls table parsing.
Type: fix
Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: Ife68c0ddc3a6f9437a149b308310f042799c2116
|
|
Type: improvement
Change-Id: I29346c849a5e1ff3c2ea399671f9f50d075e9f18
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
Type: refactor
Change-Id: I077110e1a422722e20aa546a6f3224c06ab0cde5
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: refactor
Change-Id: Ie67dc579e88132ddb1ee4a34cb69f96920101772
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
reduce the compile time by moving the bihash includes out of ip[46].h
Signed-off-by: Neale Ranns <nranns@cisco.com>
Change-Id: I6b9216e10aff1013071f9238b3e1ebbdd205bd80
|
|
Type: fix
adding routes should be MP safe. When new prefixes with differrent
prefix lengths are added, adjust the sorted list in an MP safe way.
Change-Id: Ib73a3c84d01eb86d17f8e79ea2bd2505dd9afb3d
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Type: fix
Some debugging declarations were wrapped in an 'ifdef CLIB_DEBUG'.
This seems to always evaluate true because clib.h defines CLIB_DEBUG
to 0 if its not defined.
The result is that if a route table is added and a route is added to
it and then the table is deleted, VPP exits because the debug function
fib_table_assert_empty() gets called whether VPP was started using a
debug build or a release build.
Change the ifdef to 'if CLIB_DEBUG > 0'.
Change-Id: I357dc2c299e81b95244f2f7efaadb8e0de27627a
Signed-off-by: Matthew Smith <mgsmith@netgate.com>
|
|
Type: feature
- ip-neighbour: generic neighbour handling; APIs, DBs, event handling,
aging
- arp: ARP protocol implementation
- ip6-nd; IPv6 neighbor discovery implementation; separate ND,
MLD, RA
- ip6-link; manage link-local addresses
- l2-arp-term; events separated from IP neighbours, since they are not
the same.
vnet retains just enough education to perform ND/ARP packet
construction.
arp and ip6-nd to be moved to plugins soon.
Change-Id: I88dedd0006b299344f4c7024a0aa5baa6b9a8bbe
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Type: feature
Change-Id: I28f7a658be3f3beec9ea32635b60d1d3a10d9b06
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Type: feature
the fib_source_t enum alone no longer defines the priority and
behaviour, instead each source must be allocated these attributes.
This allows the creation of other sources by the plugins (and
soon over the API).
Signed-off-by: Neale Ranns <nranns@cisco.com>
Change-Id: I890ee820fbc16079ee417ea1fbc163192806e853
|
|
Type: feature
from the API doc, a table replace is:
"
The use-case is that, for some unspecified reason, the control plane
has a very different set of entries it wants in the table than VPP
currently has. The CP would thus like to 'replace' VPP's current table
only by specifying what the new set of entries shall be, i.e. it is not
going to delete anything that already eixts.
the CP delcartes the start of this procedure with this begin_replace
API Call, and when it has populated all the entries it wants, it calls
the below end_replace API. From this point on it is of coursce free
to add and delete entries as usual.
The underlying mechanism by which VPP implements this replace is
purposefully left unspecified.
"
In the FIB, the algorithm is implemented using mark and sweep.
Algorithm goes:
1) replace_begin: this marks all the entries in that table as 'stale'
2) download all the entries that should be in this table
- this clears the stale flag on those entries
3) signal the table converged: ip_table_replace_end
- this removes all entries that are still stale
this procedure can be used when an agent first connects to VPP,
as an alternative to dump and diff state reconciliation.
Change-Id: I168edec10cf7670866076b129ebfe6149ea8222e
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Type: refactor
Signed-off-by: Simon Zhang <yuwei1.zhang@intel.com>
Change-Id: I8674ff5f6f6336b256b7df8187afbb36ddef71fb
|
|
and document scaling
Change-Id: I65d8999e65616d77e525963c770d91e9b0d5e593
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: I7a48890c075826fbd8c75436dfdc5ffff230a693
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: Ied34720ca5a6e6e717eea4e86003e854031b6eab
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Move the binary api segment above 4gb
Change-Id: I40e8aa7a97722a32397f5a538b5ff8344c50d408
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Object sizes must evenly divide alignment requests, or vice
versa. Otherwise, only the first object will be aligned as
requested.
Three choices: add CLIB_CACHE_LINE_ALIGN_MARK(align_me) at
the end of structures, manually pad to an even divisor or multiple of
the alignment request, or use plain vectors/pools.
static assert for enforcement.
Change-Id: I41aa6ff1a58267301d32aaf4b9cd24678ac1c147
Signed-off-by: Dave Barach <dbarach@cisco.com>
|
|
- IPv6 link local table is a per-SW interface array of IPv6 unicast FIBs
- the per-interface ocst is sizeof(fib_table_t) which is small,
w.r.t. the cost of an interface
- FE80::/10 in the 'global' table points to a DPO that performs a lookup in the
input interface's LL fib.
Change-Id: Ice834b25ebeeacb2e929d7c864d7ec8c09918cbe
Signed-off-by: Neale Ranns <neale.ranns@cisco.com>
|
|
Directly allocate and carve cache-line-aligned chunks of virtual
memory. To a first approximation, bihash wasn't using
clib_mem_free(...).
We eliminate mheap object header/trailers, which improves space
efficiency. We also eliminate the 4gb bihash table size limit. An 8_8
bihash w/ 100 million random entries uses 3.8 Gbytes.
Change-Id: Icf925fdf99bce7d6ac407ac4edd30560b8f04808
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
forwarding provided by the source is pushed to all other entries
it covers in the sub-tree
Change-Id: I2a45222ef653358f55c2436de3e3c6353cfadba2
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
DBGvpp# sh fib mem
FIB memory
Tables:
SAFI Number Bytes
IPv4 unicast 2 673066
IPv6 unicast 2 1054608
MPLS 1 4194312
IPv4 multicast 2 2322
IPv6 multicast 2 ???
Nodes:
Name Size in-use /allocated totals
Entry 96 20 / 20 1920/1920
Entry Source 32 0 / 0 0/0
Entry Path-Extensions 60 0 / 0 0/0
multicast-Entry 192 12 / 12 2304/2304
Path-list 40 28 / 28 1120/1120
uRPF-list 16 20 / 20 320/320
Path 72 28 / 28 2016/2016
Node-list elements 20 28 / 28 560/560
Node-list heads 8 30 / 30 240/240
Change-Id: I8c8f6f1c87502a40265bf4f302d0daef111a4a4e
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: I4b4648831551519b2ffb6f93255d28a4b8726c22
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
part 2;
- this adds the code to create an IP and MPLS table via the API.
- but the enforcement that the table must be created before it is used is still missing, this is so that CSIT can pass.
Change-Id: Id124d884ade6cb7da947225200e3bb193454c555
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
According to Maciek, the easiest way to leverage the csit "performance
trend" job is to actually merge the patch once verified. Manual
testing indicates that the patch improves l2 path performance. Other
use-cases are TBD. It's possible that we'll need to back out the patch
depending on what happens.
Change-Id: Ic0a0363de35ef9be953ad7709c57c3936b73fd5a
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
- inline the FIB lookup function; this requires access to the bihash, so for files that use more than one type this casues problems. those files that include ip6_fib.h unnecessarily have been updated
- better use of the feature arcs. ip6-lookup and interface-output are now sentinels (end-node-index in the cm speak) rather than enabled features.
Change-Id: I9d1375fee63f7dbb2d327da6124d8e60b63367ec
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: Iae04c57bba87ab3665388eadd0805f75171636a5
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
- the flow hash config is (and was) cached on the load-balance object so the fib_table_t struct is not used a switch time. Therefore changes to the table's flow hash config need to be propagated to all load-balances and hance all FIB entries in the table.
- enable API for setting the IPv6 table flow hash config
- use only the hash config in the fib_table_t object and not on the ipX_fib_t
- add tests.
Change-Id: Ib804c11162c6d4972c764957562c372f663e05d4
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
tests). The DPO was incorrectly initialised with FIB_PROTO_MAX
Change-Id: I962df9e162e4dfb6837a5ce79ea795d5ff2d7315
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
- single-hop BFD: attach a delegate to the appropriate adjacency
- multi-hop BFD [not supported yet]: attach a delegate to the FIB entry.
adjacency/fib_entry state tracks the BFD session state. when the state is down the object does not contribute forwarding hence and hence dependent objects will not use it.
For example, if a route is ECMP via two adjacencies and one of them is BFD down, then only the other is used to forward (i.e. we don't drop half the traffic).
Change-Id: I0ef53e20e73b067001a132cd0a3045408811a822
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
1) 16-8-8 stride. Reduce trie depth walk traded with increased memory in the top PLY.
2) separate the vector of protocol-independent (PI) fib_table_t with the vector of protocol dependent (PD) FIBs. PD FIBs are large structures, we don't want to burn the memory for ech PD type
3) Go straight to the PD FIB in the data-path thus avoiding an indirection through, e.g., a PLY pool.
Change-Id: I800d1ed0b2049040d5da95213f3ed6b12bdd78b7
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
- IPv[46] mfib tables with support for (*,G/m), (*,G) and (S,G) exact and longest prefix match
- Replication represented via a new replicate DPO.
- RPF configuration and data-plane checking
- data-plane signals sent to listening control planes.
The functions of multicast forwarding entries differ from their unicast conterparts, so we introduce a new mfib_table_t and mfib_entry_t objects. However, we re-use the fib_path_list to resolve and build the entry's output list. the fib_path_list provides the service to construct a replicate DPO for multicast.
'make tests' is added to with two new suites; TEST=mfib, this is invocation of the CLI command 'test mfib' which deals with many path add/remove, flag set/unset scenarios, TEST=ip-mcast, data-plane forwarding tests.
Updated applications to use the new MIFB functions;
- IPv6 NS/RA.
- DHCPv6
unit tests for these are undated accordingly.
Change-Id: I49ec37b01f1b170335a5697541c8fd30e6d3a961
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: I7b51f88292e057c6443b12224486f2d0c9f8ae23
Signed-off-by: Damjan Marion <damarion@cisco.com>
|