summaryrefslogtreecommitdiffstats
path: root/src/vnet/l2/l2_input.c
AgeCommit message (Collapse)AuthorFilesLines
2020-09-14l2: allocate l2fib only when neededDamjan Marion1-0/+3
Currently l2 fib allocates 512MB hash table unconditionally on startup. This patch postpones table creation up to the point where first interface is put into l2 mode or mac entry is added. In addition it reduces default table size to 128MB and increases number of buckets 4 times. This default setting should be enough to keep 1M mac entries. Also, new startup.conf section is added which allows user to change memory and bucket size. .i.e: l2fib { table-size 512M num-buckets 524288 } Type: improvement Change-Id: I2a29209aa3545181f0087544c97a54d8157b6ec5 Signed-off-by: Damjan Marion <damarion@cisco.com>
2020-06-27l2: performance enhancement in l2inputZhiyong Yang1-80/+69
Short Load/Stores combined with prefetching in the beginning of the loop place too much pressure on AGUs and memory accesses. The patch interleaves load/store operations with computational operations to alleviate the pain point. vlib_get_buffers is also leveraged. Redefine u8 dst_and_src[12] instead of dst[6] and src[6] in struct l2input_trace_t in order to merge two copys into one. Type: improvement Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com> Change-Id: I7d3df7732c476069235e3019c68f0f53bca9637e
2020-05-16l2: L2/L3 mode swicth cleanup and l2-patch fixJohn Lo1-16/+6
Cleanup L2/L3 mode switch to not redirect to/from ethernet-input node as it is no longer necessary. L2 patch should use sw_if_index for device feature enable/disable. Type: fix Signed-off-by: John Lo <loj@cisco.com> Change-Id: I0f24161d027b07c188fd1e05276146f94c075710
2019-12-17l2: fix l2input_feat_names overflowBenoît Ganne1-1/+1
Type: fix Change-Id: I59549b3aab5a0ccfe7db02757f78528e028121a5 Signed-off-by: Benoît Ganne <bganne@cisco.com>
2019-04-24l2: Add support for arp unicast forwardingMohsin Kazmi1-1/+4
Change-Id: I79fc55f36a9b83957f84619bdf8cef08acc8ec24 Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
2019-03-12l2: migrate old MULTIARCH macros to VLIB_NODE_FNFilip Tehlar1-6/+15
Change-Id: I8370c06150ce4499475e9d6dc6b3ab8be2016202 Signed-off-by: Filip Tehlar <ftehlar@cisco.com>
2018-11-14Remove c-11 memcpy checks from perf-critical codeDave Barach1-10/+10
Change-Id: Id4f37f5d4a03160572954a416efa1ef9b3d79ad1 Signed-off-by: Dave Barach <dave@barachs.net>
2018-11-13L2 feautre bitmaps output verbose/non-verbose modeNeale Ranns1-2/+11
Change-Id: I15ff191ee8724a3354c074db590472db05e0652e Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-10-24L2-input/output: use feature enum type in flag update functionNeale Ranns1-1/+2
Change-Id: I1f58f441c65fbca101bee2e864bfa6ae2306b475 Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-10-23c11 safe string handling supportDave Barach1-1/+1
Change-Id: Ied34720ca5a6e6e717eea4e86003e854031b6eab Signed-off-by: Dave Barach <dave@barachs.net>
2018-10-15vnet: create L2 feature arc infrastructureAndrew Yourtchenko1-1/+1
This commit adds two new nodes in the L2 datapath in input and output direction respectively. These nodes fork the traffic into three feature arcs: ip4, ip6 and nonip, which later join to continue the usual L2 processing. The vnet_l2_feature_enable_disable() function with the same signature as vnet_feature_enable_disable() takes care of enabling the L2 datapath feature bits as needed, when the features are enabled/disabled. Thus, L2 features may use the similar plumbing as the L3 features enjoy. Change-Id: I76877b3a92d794c492bff1622bb26acba05705b2 Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
2018-10-11L2: BM traffic does not use UU-FWD (VPP-1445)Neale Ranns1-1/+2
Change-Id: I5495c37da2fb8ff48c4af14ccba021d64eac52b6 Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-09-25L2 BD: introduce a BD interface on which to send UU packetsNeale Ranns1-19/+49
Change-Id: I21ad6b04c19c8735d057174b1f260a59f2812241 Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-09-24Trivial: Clean up some typos.Paul Vinciguerra1-4/+4
Change-Id: I085615fde1f966490f30ed5d32017b8b088cfd59 Signed-off-by: Paul Vinciguerra <pvinci@vinciconsulting.com>
2018-09-14Revert "L2-input: use vlib_buffer_enqueue_to_next"John Lo1-317/+239
The patch did not improve single input link performance and degrade performance with multiple input links. This reverts commit b1232555e91b286feab5667b5a22f29aa8e96626. Change-Id: Ib9336a2e0610088b9145a43cacbdadb52dabd522 Signed-off-by: John Lo <loj@cisco.com>
2018-09-13L2-input: use vlib_buffer_enqueue_to_nextNeale Ranns1-239/+317
use the same trick as l2-ouput to group the processing of packets on the same interface. Change-Id: Ib2a6a1b5f362372936197f5bb2fdd0fe9439226b Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-09-08L2 BVI/FIB: Update L2 FIB table when BVI's MAC changesNeale Ranns1-0/+21
also some moving of l2 headers to reduce dependencies Change-Id: I7a700a411a91451ef13fd65f9c90de2432b793bb Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-09-07L2-FIB: replace bit-fields with flagsNeale Ranns1-1/+3
Change-Id: Ic31da442a0e0477569d53b4a72627bbb25e93365 Signed-off-by: Neale Ranns <nranns@cisco.com>
2018-07-15VPP-1341: fix loopback interface graph arcsDave Barach1-15/+0
Remove broken special case from l2_input.c:set_int_l2_mode(), which turns out to confuse the graph dispatch engine. The loopback TX function needs to push packets to either ethernet-input or to l2-input, based on bridge / BVI configuration. Rather than overloading a single graph arc - and making vain attempts to reconfigure it - create both arcs and use the correct one. Rewrote the loopback tx function as an idosyncratic multi-arch quad/single loop fn. Change-Id: I15b56ce641d90a11e7b3c7d23859f40e168dd7b2 Signed-off-by: Dave Barach <dave@barachs.net>
2018-05-31Improve L2FIB delete entry handling and "show l2fib" CLI optionsJohn Lo1-1/+1
Improve deletion of L2FIB MAC entry using l2fib_add_del API. If sw_if_index param specified in th API is non-zero, check that its value match that of the MAC entry to proceed with deletion. Improve "show l2fib" CLI to allow display of all entries, learned entries only, or added ovia CLI/API entries only. For added entry, show "no" under the age column to indicate entry does not age. Change-Id: I0bd2582c2b6bac268e551e4f8ca6dab2be4400ad Signed-off-by: John Lo <loj@cisco.com>
2018-04-13GBP V2Neale Ranns1-1/+2
update the GBP plugin to implement the full NAT feature set of opflex agent Change-Id: Ic06a039c889445ed0b9087fa1f292634192b0f8d Signed-off-by: Neale Ranns <neale.ranns@cisco.com>
2018-04-13bond: ping fails between l2 BD [VPP-1238]Steven1-0/+7
In dpdk based bonding, when the bond interface is configured for l2, it automatically sets the bond interface to promiscuous mode and sets rx redirect to ethernet-input. This allows traffic to be bridged to non compute node facing interface when it is received from the compute node interface. For native vpp bonding, we need to do similar things. When the bond interface is configured for l2, we set the slave interfaces to promiscuous mode and set rx redirect to ethernet-input because dpdk does not know anything about the bond interface. Likewise, when a new interface is enslaved, we also need to do the same thing if the bond interface has already been configured for l2. Change-Id: I7e168008e8a4221be74929b2a20e6db0ce8f3110 Signed-off-by: Steven <sluong@cisco.com>
2018-04-09L2: no-flood interface type in the Bridge-DomainNeale Ranns1-0/+12
Change-Id: I50ff0cacf88182f8e0be19840c50f4954de586e2 Signed-off-by: Neale Ranns <neale.ranns@cisco.com>
2018-03-29l2_input:optimize counter accessEyal Bari1-31/+16
only one counter update per frame (was updated per iteration) only access ethertype for casts (was always accessing ethertype) Change-Id: I3a3c3219ec63e975cf5bd8cf2d93103932a4aaa3 Signed-off-by: Eyal Bari <ebari@cisco.com>
2018-02-07classifier-based ACL: refactor + add output ACLAndrew Yourtchenko1-2/+1
For implementation of MACIP ACLs enhancement (VPP-1088), an outbound classifier-based ACL would be needed. There was an existing incomplete code for outbound ACLs, it looked almost exact copy of input ACLs, minus the various enhancements, trying to sync that code seemed error-prone and cumbersome to maintain in the longer run. This change refactors the input+output ACLs processing into a unified routine (thus any changes will have effect on both), and also adds the API to set the output interface ACL, with the same format and semantics as the existing input one (except working on output ACL of course). WARNING: IP outbound ACL in L3 mode clobbers the ip.* fields in the vnet_buffer_opaque_t, since the code is using l2_classify.* The net_buffer (p0)->ip.save_rewrite_length is rescued into l2_classify.pad.l2_len, and used to rewind the header in case of drop, so that ipX_drop prints something sensible. Change-Id: I62f814f1e3650e504474a3a5359edb8a0a8836ed Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
2017-11-10Further fix to SHG handling for ARP/ICMPv6 from BVI in a BDJohn Lo1-6/+25
For ARP/ICMPv6 packets received from a BVI in a BD, allow flood to all remote VTEPs via VXLAN tunnels irrespective of SHG check for ARP request or ICMPv6 neighbor solicitation packets only. All other packets types will flood normally as per SHG check. Change-Id: I17b1cef9015e363fb684c2b6506ed6c4efe70bba Signed-off-by: John Lo <loj@cisco.com> (cherry picked from commit 5b99133cff1ff0eb9043dd8bd3648b0b3aafa47e)
2017-11-08Fix SHG handling for ARP/ICMPv6 received from BVI in a BDJohn Lo1-0/+8
This change makes sure ARP/ICMPv6 brodcast packets received from the BVI of a BD can be flooded to all remote VTEPs via its VXLAN tunnels irrespective of SHG setting. Similar processing was done for unicast packets already and needs to be extpanded to ARP and ICMPv6 broadcast packets. Change-Id: I26ac43ecdbc81a769f742a583a156506f7e70d49 Signed-off-by: John Lo <loj@cisco.com> (cherry picked from commit c97b4aca0db8d84b17ceb03a14ab44346a2b3466)
2017-10-31l2fib: MAC: Fix uint64 to u8 byte arrayMohsin Kazmi1-5/+2
As per proposal on the mailing list, this patch fixes the represntation of MAC address in VPP API calls for · L2fib_add_del · L2_fib_table_details Change-Id: I31e17efd1a6314cded69666e693cb8fc33158d02 Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
2017-10-03Update L2FIB entry timestamp only if BD aging enabled (VPP-1002)John Lo1-0/+1
Change L2 learning path so it update stale timestamp in MAC entry only if aging is enabled on the BD for the MAC entry. Change-Id: I7babe986ceef3c030d8ef9185076c42b405f7b0f Signed-off-by: John Lo <loj@cisco.com>
2017-07-31CLI:add l2 input/outut to "sh int features"Eyal Bari1-0/+23
Change-Id: If608bbc7f4c8b0d5c3a237098a20279e407c82d3 Signed-off-by: Eyal Bari <ebari@cisco.com>
2017-07-23Improve L2 Input/Output Feature Infrastructure and UsageJohn Lo1-20/+17
Simplify L2 output feature infra to unify with L2 input feature infra using the newly improved feature bitmap mechanism. Updated all L2 features to use the more efficient infra functions. Change-Id: If8f463826b0af0717129befe92a27ea8cfc40449 Signed-off-by: John Lo <loj@cisco.com>
2017-07-20L2FWD:move vec_validate out of access functionEyal Bari1-1/+1
Change-Id: Id9737b6aa2b6fe3032f4627dfdbd2ea728cc3fb1 Signed-off-by: Eyal Bari <ebari@cisco.com>
2017-07-13Fix crash with worker threads on 4K VXLAN/BD setup (VPP-907)John Lo1-25/+24
Cleanup mapping of interface output node for the l2-output node when interface is configured to L2 or L3 modes. The mapping is now always done in the main thread as part of API/CLI processing, instead of initiate mapping in the forwarding path which can be in the worker threads. Change-Id: Ia789493e7d9f5c76d68edfaf34db43f3e3f53506 Signed-off-by: John Lo <loj@cisco.com> (cherry picked from commit bea5ebf205e0bec922bf26c6c1a6a9392b4cad67)
2017-06-27L2-LEARN:fix l2fib entry seq num not updated on hit (VPP-888)Eyal Bari1-1/+1
fixed instability in l2bd_multi_instnce test - sometimes failing with extra packets captured it appears l2-learn was not updating hit entries but rather a copy of them. if the ager did not have a chance to run before the test was running the learning cycle - entries were not updated with the packet's seq num - causing packets to flood when hitting the stale seq_num in l2-fwd - hence the extra packets fixed handling of filter entries revert workaround for instability in test Change-Id: I16d918e6310a5bf40bad5b7335b2140c2867cb71 Signed-off-by: Eyal Bari <ebari@cisco.com> (cherry picked from commit 25ff2ea3a31e422094f6d91eab46222a29a77c4b)
2017-06-19L2FWD:fix seq_num overwritten + validate l2fib entries when forwardingEyal Bari1-2/+2
l2_classify memeber table_index was overlaid over l2.l2fib_seq_num which over written when table_index gets initialized in l2_input_classify solved by overlaying both table_index and opaque_index as only one is used seperated l2fib seq num from l2_input configs for better handling of theoretical ABA issue where an entry for a deleted interface is considered valid by the ager because a different interface with same sw_if_index and seq_num was created before the ager got a chance to delete Change-Id: I7b0eeded971627406f1c80834d7e02c0ebe62136 Signed-off-by: Eyal Bari <ebari@cisco.com>
2017-05-19Enforce Bridge Domain ID range to match 24-bit VNI rangeJohn Lo1-0/+6
Enforce bridge domain ID range to allow a maximum value of 16M which matches the range of 24-bit VNI used for virtual overlay network ID. Fix "show bridge-domain" output to allow full 16M BD ID range to be displayed using 8-digit spaces. Change-Id: I80d9c76ea7c001bcccd3c19df1f3e55d2970f01c Signed-off-by: John Lo <loj@cisco.com>
2017-05-08L2FIB:CLI/API to flush all non-static entriesEyal Bari1-5/+8
added CLI l2fib flush-mac all added API l2fib_flush_all flushes all non static l2fib entries on all valid BDs Change-Id: Ic963c88f4bed56308c03ab43106033132a0e87be Signed-off-by: Eyal Bari <ebari@cisco.com>
2017-05-03L2FIB:flush interface learned macs on downEyal Bari1-28/+14
Change-Id: I80a723f55fcf2ecc3209a35e8297c88b45b1abfb Signed-off-by: Eyal Bari <ebari@cisco.com>
2017-04-06Use thread local storage for thread indexDamjan Marion1-7/+7
This patch deprecates stack-based thread identification, Also removes requirement that thread stacks are adjacent. Finally, possibly annoying for some folks, it renames all occurences of cpu_index and cpu_number with thread index. Using word "cpu" is misleading here as thread can be migrated ti different CPU, and also it is not related to linux cpu index. Change-Id: I68cdaf661e701d2336fc953dcb9978d10a70f7c1 Signed-off-by: Damjan Marion <damarion@cisco.com>
2017-03-28Implement MAC Flush for BD or Interface from the L2FIBJohn Lo1-0/+14
Allow non-static MACs in the L2FIB which is associated with an interface or a bridge domain (BD) be flushed. MAC flush are initiated automatically when an interface is removed from a BD or when a BD is deleted. MAC flush can also be invoked manually via the following CLI: l2fib mac-flush interface <if-name> l2fib mac-flush bridge-domain <bd-id> Change-Id: Ie33243622834810a765f48ebcd22bdb8e8fc87a4 Signed-off-by: John Lo <loj@cisco.com>
2017-03-03VPP-651: Ensure sw_if_index to node mapping for L2 output path is only done ↵Andrew Yourtchenko1-1/+2
via l2output_main.next_nodes Before this commit, several output features that happen to be the last in the list of features to be executed, send the packets directly to <interfaceName>-output. To do this, they use l2_output_dispatch, which builds a list of sw_if_index to next index mappings. When interfaces are deleted and the new interfaces are created, these mappings become stale, and cause the packets being sent to wrong interface output nodes. This patch (thanks John Lo for the brilliant idea!) adds a feature node "output", whose sole purpose is dispatching the packets to the correct interface output nodes. To do that, it uses the l2output_main.next_nodes, which is already taken care of for the case of the sw_if_index reuse, so this makes the dependent features all work correctly. Since this changes the packet path, for the features that were always the last ones it has triggered a side problem of the output feat_next_node_index not being properly initalized. These two users are l2-output-classify node and the output nodes belonging to the acl-plugin. For the first one the less invasive fix is just to initialize that field. For the acl-plugin nodes, rewrite the affected part of the code to use feat_bitmap_get_next_node_index since this is essentially what the conditional in l2_output_dispatch does, and fix the compiler warnings generated. This fix was first made in stable/1701 under commit e7dcee4027854b0ad076101471afdfff67eb9011. Change-Id: I32e876ab1e1d498cf0854c19c6318dcf59a93805 Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
2017-02-28Clear L2 output config on interface mode change to L3 (VPP-651)John Lo1-4/+11
With VPP-651, the L2 output config with L2-tag rewrite was not cleared when a sub-interface is deleted. Subsequently, when the same sw_if_index was reused for another interface, the L2 output config with L2-tag rewrite remained on the new interface. On deleting a (sub-)interface which is in L2 mode, it will be changed to L3 mode first to clear any L2 config. The L2 to L3 mode change path did address L2 input config cleanup. It is now fixed to also clear L2 output config. Change-Id: I3352a89d92e1b27340a5adcf75bbaa01a5050c29 Signed-off-by: John Lo <loj@cisco.com>
2017-02-17l2 input: avoid per-packet trace checks in the fast pathDave Barach1-7/+14
Change-Id: Ib0c8572773499d8dd4d81b3a565c24412ccc3510 Signed-off-by: Dave Barach <dave@barachs.net>
2016-12-28Reorganize source tree to use single autotools instanceDamjan Marion1-0/+1116
Change-Id: I7b51f88292e057c6443b12224486f2d0c9f8ae23 Signed-off-by: Damjan Marion <damarion@cisco.com>