Age | Commit message (Collapse) | Author | Files | Lines |
|
The patch brings 0.8 clocks saved per pkt in IPv4 l3fwd case on Skylake.
Type: improvement
Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
Change-Id: Ia8d3a27773bf959433380d7c219602b1e4a8e5bd
|
|
It is a bad idea to poison memory after munmap because the address space
can be reused (eg. for global data of dlopen()ed object) and ASan model
allows access by default.
Moreover, access to a stale address space will fault.
Type: fix
Change-Id: I356de422f255447d9d50a3a71fb0c2eaa790d731
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: Ia04bd26ecd689894753e036e52920316de611910
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Measured improvement is from 439 to 167 clocks for add operation
in 16_8 case...
Type: improvement
Change-Id: I975ff46ff30b983a3ec80a5cde25ccb68d7fa03b
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
0001 ~ 0014 patches are for virtual channel and PMD
0015 is the iavf fdir framework
0016 ~ 0017 are for the iavf fidr driver
Type: feature
Signed-off-by: Chenmin Sun <chenmin.sun@intel.com>
Change-Id: I38e69ca0065a71cc6ba0b44ef7c7db51193a0899
|
|
Type: fix
Change-Id: I3c7ebbe91e7dffe9fd6851e5334fe920f2187cf0
Signed-off-by: Amir Zeidner <amirzei@mellanox.com>
|
|
vl_api_address_t uses a packed enum for the address family, compare
a->af directly with ADDRESS_IP4 / ADDRESS_IP6 instead of running a->af
through clib_host_to_net_u32(...) before compare.
Indirectly fixes api_ip_route_add_del(...) w/ count > 1.
Type: fix
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: Ib7f562ec9e92ee63e52a338d318fcf1ce6221755
|
|
Adapt queue_hi_thresh value using num_threads to avoid risk of deadlock
between threads which could happen for example when different NAT
threads try to handoff work to each other at the same time when their
frame queues are congested. This change ensures that each thread can
reserve a queue entry without causing problems even in the most extreme
case when all threads attempt to add to the same queue simultaneously
when the queue is nearly full.
Type: fix
Signed-off-by: Elias Rudberg <elias.rudberg@bahnhof.net>
Change-Id: I9e02f753bd00833d8dd500d181b0d4f9a454d703
|
|
Previously dslite was moved to separate plugin folder
and CE mode was removed. But it still needed.
This patch adds CE option to separate config entry
Type: feature
Signed-off-by: Vladimir Ratnikov <vratnikov@netgate.com>
Change-Id: If153ae08fa385ba5a6605cb412e49bbb4d1db46c
|
|
Template instances can allocate BIHASH_KVP_PER_PAGE data records
tangent to the bucket, to remove a dependent read / prefetch.
Template instances can ask for immediate memory allocation, to avoid
several branches in the lookup path.
Clean up l2 fib, gpb plugin codes: use clib_bihash_get_bucket(...)
Use hugepages for bihash allocation arenas
Type: improvement
Signed-off-by: Dave Barach <dave@barachs.net>
Signed-off-by: Damjan Marion <damarion@cisco.com>
Change-Id: I92fc11bc58e48d84e2d61f44580916dd1c56361c
|
|
Type: fix
Change-Id: Ia5ae0e893a5358f61353d20f444d88d79953e482
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Signed-off-by: Benoît Ganne <bganne@cisco.com>
Change-Id: Ib85c2f01dc7ec9858f2f88b89e209f989d78c5d9
|
|
Type: fix
Change-Id: Ie60b07abe76ad166f048f5885accd7038d8153b2
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Iba37e528e968104c3ba9c8324438ba695ddddfd1
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I4d56b61af574dfdaf6028160ce331606bdf65609
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I56c487821233cebf2146745a4706cb573cc088a5
|
|
Type: improvement
Signed-off-by: Klement Sekera <ksekera@cisco.com>
Change-Id: If5c00faa309cf3e1ea8bdf8b23250041f6a499c4
|
|
Type: fix
Change-Id: Ib455b0a57f9b4f9cb82bb295c220270d0c6e5fe5
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Signed-off-by: Simon Zhang <yuwei1.zhang@intel.com>
Change-Id: Id84fbe412c99b39a0754b892ca971dd3e4434264
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I931d23cf617c8f24fe9a02982ab129785bbd184d
|
|
Type: fix
Change-Id: I42c3e3514ba50d40e09ee688d083f3e78fa0713a
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
replace hand crafted version with existing vector function
Type: fix
Signed-off-by: Andreas Schultz <andreas.schultz@travelping.com>
Change-Id: I94834852469ed316c7bdbd48bb7f479648b2c8e5
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I48ce827966667a767440149769eb1b24616ce4c1
|
|
Some vhost-backed missing the VIRTIO_RING_F_INDIRECT_DESC feature.
Previously, vpp doesn't support jumbo frames using chained descriptors.
This patch fixes this issue.
Type: fix
Change-Id: I20487e201e88ea136b556ac84dde058019ab3e78
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Now UDP encapsulation doesn't work in transport mode because:
- the encrypt node misses filling of UDP header and it gets sent with
all zeros;
- the decrypt node misses filling of new IP header and it contains
garbage data.
With this commit, fill UDP header during encryption and fill IP header
during decryption.
Change-Id: I87a7bd594f0e312b16d3e5eb19e568b4e3164d36
Type: fix
Signed-off-by: Alexander Chernavin <achernavin@netgate.com>
|
|
Type: improvement
Do extra checks when establishing an active connect and cleanup pending
connects if application detaches.
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ibe9349db57b313ba2aa5ea3960ef5cf755f5098a
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I21daa87b1a301142270a1bf20680dd9e9395aac9
|
|
Type: improvement
Signed-off-by: Yu Sun <yusun2@cisco.com>
Change-Id: I68aea7c5776c5b31081c98388df4133d2062218a
|
|
Type: feature
Change-Id: I699a01ac925fe5c475a36032edb7018618bb4dd4
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Type: fix
Change-Id: Ie7081d977dd0d3e7d09bc0d1b4d53863288e443b
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Change-Id: I1068ff1d03883addb5fe1005a8b608eab17e4168
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
The vector is initialized to 1024 entries which is guaranteed to be
enough, but as its size can shrink between calls, make sure ASan is
aware of the expected size before using it.
Type: fix
Change-Id: I4bcc39867a886b3cb463854d2cda0b32155650e9
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
When merging proxy groups in igmp_proxy_device_merge_group(), the call
to igmp_proxy_device_merge_src() can end up removing the current proxy
group via igmp_group_clear(). When that happens, it must returns NULL so
that igmp_proxy_device_merge_config() does not send a IGMPv3 report for
a dead proxy group.
Make igmp_group_clear() reset the group pointer to NULL to fix this bug
and to detect similar bugs more easily.
Type: fix
Change-Id: I229e55b5bfa71734d7844893f5209a66fa3cc8ae
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Signed-off-by: Andreas Schultz <andreas.schultz@travelping.com>
Change-Id: I774ca50582281cb3cc5e43417e74d178bf4909bd
|
|
Type: refactor
Change-Id: I9f743ba2818e1b1c5004c3575925cc7b479948d8
Signed-off-by: Filip Varga <fivarga@cisco.com>
|
|
Type: improvement
Change-Id: I7e11bf72be5fad5967724c038eb649a261294ca0
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
There is no need to calculate bucket2 if there is hit on bucket1
Type: improvement
Change-Id: Id01c37963497668c0160068501294568a181d011
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I547263ae954506f11101666ff768524fbfdb579e
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
See .../extras/pg/checksum_offload.pg for a nontrivial example, which
deliberately sets bogus ip and udp checksums in the generated packets,
then fixes the mess with (software emulated) hardware checksum
offload.
Validated via "pcap dispatch trace on max 1000 buffer-trace pg-input 100".
Packets stuffed into loop1-output have the configured bogus ip and udp
checksums. vnet_calc_checksums_inline(...) fixes the checksums, which
are correct when packets visit loop1-tx.
The packet generator is a dumb robot in this regard. If you ask for a
ridiculous flag combination - example: ip4 and ip6 - your mileage may
vary.
Type: feature
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I6d9e790f373bcd9e445a89113ca8e4c8f9aa9419
|
|
Type: improvement
Change-Id: Ifb4eec00fd4f1d19e4b0af802d015a35e402e0af
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Avoids unwanted notifications.
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ifff27fb0d3644194e3bb3f91d7ed1fd02c6730cd
|
|
Type: fix
Change-Id: I87d301aec20b9f5b34997b394493c796188fce14
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Id0c4b46247e7419a328387678a4753cfb1e42d75
|
|
Merge two mildly incompatible static inlines, and rename the results
vnet_calc_checksums_inline (...).
The resulting inline has three additional parameters: int is_ip4, int
is_ip6, and int with_gso. All calls manage to pass one or more as
compile-time constants, which causes a certain amount of code to
disappear in each instantiation.
Type: refactor
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I1a2a0e353b9a7bca20bc81318e8c915617261e1a
|
|
vat_socket_connect(...) needs to set the api main client index. The M2
helper macro needed to allocate a buffer of size sizeof(*mp) + n, not
sizeof(*mp).
Type: fix
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I7abf35d2ba045d82765bd27f88899287fd351602
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ifff49366ad538bf2dd64487cae17e7457dff064f
|
|
Type: feature
- move the IP4 code to plugin
- add ip6 support
- add suport for uRPF on TX
- add tests
Change-Id: I074c2debc486d3e79c12fad4b8dbd72c41e841a0
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Type: fix
There is a statically allocated array for inbound SAs which can hold
4 IDs. The input parameter containing the IDs of th inbound SAs is a
vector and Its possible to pass a vector with more than 4 elements
and write the memory past the end of the array. Fail if more than 4
SAs are passed in the vector.
Change-Id: I0c9d321c902d6366b8aff816d04e343dcbd110eb
Signed-off-by: Matthew Smith <mgsmith@netgate.com>
|
|
Type: refactor
UDPC (removed) was the only consumer.
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I2812bf58d3c68fe021ec73acaa1bd00ef3172846
|
|
When configured at compile time via the cmake
VPP_BUFFER_FAULT_INJECTOR option, the buffer allocator will appear to
fail a certain fraction of the time.
By default, the allocator succeeds 80% of the time. Detailed command
line configuration options are available, but only when the image has
been compiled with cmake option described above:
vlib { buffer-alloc-success-rate [0.0 ... 1.0]
buffer-alloc-success-seed <nnnn> }
Modify vlib_buffer_pool_create(...) so 0 is always an invalid buffer
index.
Debug images: add checks for bad buffer index enqueues, and also
verify that f->n_vectors doesn't accidentally map one or more
instances of the frame poison pattern 0xfefefefe.
Type: improvement
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: Iab939858014463d1e664682805013d334d6fcbe5
|