Age | Commit message (Collapse) | Author | Files | Lines |
|
Because the code is not optimized, newreno is still the default
congestion control algorithm.
Change-Id: I7061cc80c5a75fa8e8265901fae4ea2888e35173
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
It is not used and just confuses people...
Change-Id: Ic731432a785731271531f183b448e4591a1d2a8b
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Idd4471a3adf7023e48e85717f00c786b1dde0cca
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Idbce0393fc9e6e8dbb2765ed164ba7f90d1ffccc
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
- update pacer once per burst
- better estimate initial rtt
- compute smoothed average for higher precision rtt estimate
Change-Id: I06d41a98784cdf861bedfbee2e7d0afc0d0154ef
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Instead of reusing buffers for acking, consume all buffers and program
output for (dup)ack generation. This implicitly fixes the drop counters
that were artificially inflated by both data and feedback traffic.
Moreover, the patch also significantly reduces the ack traffic as we now
only generate an ack per frame, unless duplicate acks need to be sent.
Because of the reduced feedback traffic, a sender's rx path and a
receiver's tx path are now significantly less loaded. In particular, a
sender can overwhelm a 40Gbps NIC and generate tx drop bursts for low
rtts. Consequently, tx pacing is now enforced by default.
Change-Id: I619c29a8945bf26c093f8f9e197e3c6d5d43868e
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I30487bd736407378fb5a6d313e4eef12bbb262b8
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Learning GBP endpoints over vxlan-gbp tunnels
Change-Id: I1db9fda5a16802d9ad8b4efd4e475614f3b21502
Signed-off-by: Neale Ranns <neale.ranns@cisco.com>
|
|
Change-Id: I29f20dbaf2c2d735faff297cee552ed648f6f61b
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
(gdb) bt
bt
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb) frame 5
frame 5
293 if (PREDICT_FALSE (rxvq->last_avail_idx == rxvq->avail->idx))
(gdb) p *rxvq
p *rxvq
$3 = {cacheline0 = 0x7f290bcadd80 "\377\003", qsz_mask = 1023, last_avail_idx = 0, last_used_idx = 0, n_since_last_int = 0, desc = 0x0, avail = 0x0, used = 0x0, int_deadline = 0, started = 1 '\001', enabled = 1 '\001', log_used = 0 '\000', cacheline1 = 0x7f290bcaddc0 "\377\377\377\377\016", errfd = -1, callfd_idx = 14, kickfd_idx = 19, log_guest_addr = 5151049792, mode = 0}
The crash is because we access the null pointer rxvq->avail,
which is supposed to be derived from the mmap informed by the driver.
We fixed a similar issue before in
https://gerrit.fd.io/r/#/c/14545/
The reason was the driver ummaps the memory without doing the disconnect in
SR-IOV environment. The fixed was applied to the RX path. Now it happens in the
TX path. We just need to apply the same check in the TX path.
Change-Id: I7b1dfc96797cb5b52845bc6cec09a8c5d4325280
Signed-off-by: Steven <sluong@cisco.com>
|
|
The change can save 1.1 clocks per packet on Intel Atom C3858 platform,
It downgraded from 2.05e1 to 1.94e1 clocks per packet.
The change can save 0.3 clocks per packet on Intel Xeon CPU E5-2699 v4 @ 2.20GHz,
It downgraded from 1.26e1 to 1.23e1 clocks per packet.
Change-Id: I1ede77fb592a797d86940a8abad9ca291a89f1c7
Signed-off-by: Yulong Pei <yulong.pei@intel.com>
|
|
Avoid dequeuing acked bytes more than once per burst for a connection.
Although the fifos do not use locks, size decrements are atomic, so they
rely on locked instructions.
Change-Id: Id65f4ea40b2c10057461402dfd0393034e6472d5
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
For vxlan_encap, code will touch memory area before the field "data"
in struct vlib_buffer_t, however so far it is not prefetched in cache
yet for this graph node.
After applying the patch, 2~3 cycles per pkt for vxlan4_encap can be
saved on Haswell. It will bring a lot of benefits on DVN platform too.
Change-Id: I26d8c57fb3d2415726be5367117d73eb715e35ad
Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
|
|
Add atomic swap and store macro with acquire and release ordering
respectively. Variable in question is interupt_pending variable which
is used as guard variable by input nodes to process the device queue.
Atomic Swap is used with Acquire ordering as writes or reads following
this in program order should not be reordered before the swap.
Atomic Store is used with Release ordering, as post store the node is
added to pending list.
Change-Id: I1be49e91a15c58d0bf21ff5ba1bd37d5d7d12f7a
Original-patch-by: Damjan Marion <damarion@cisco.com>
Signed-off-by: Sirshak Das <sirshak.das@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Ola Liljedahl <ola.liljedahl@arm.com>
|
|
Change-Id: I8bc3a991f0ede0605d78b51ba609fbe5889513f2
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: Idd4a5f8bab5d39e5f33f5c130601175af70a20d4
Signed-off-by: Filip Varga <filip.varga@pantheon.tech>
|
|
Allows sending of unsent data in fast recovery and consolidates logic in
tcp, instead of splitting it between tcp fast retransmit and tcp output
path called by the session layer.
Change-Id: I9b12cdf2aa2ac50b9f25e46856fed037163501fe
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ib15d629c5fde7849bfa3307f42659e920eb0f463
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Refactor most of the ping code to be address-family agnostic,
and add support for chained buffers (thus, sending
the payloads bigger than 2K).
Change-Id: I749c302ca2f3390e0d1f84046fc72da5cf13e3ef
Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
|
|
Change-Id: I9b5f7b264f9978e3dd97b2d1eb103b7d10ac3170
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Also reset pacer on tcp retransmit timeout
Change-Id: I5a9edee4c00d1d169248d79587a9b10437c2bd87
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Also propagate tcp worker context instead of retrieving it multiple
times.
Change-Id: I7b273b981826b37783566d0172a64cd6957f3b33
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ided6c661edc9e2035fd7b472c312e2380d3f9c0b
Signed-off-by: Eyal Bari <ebari@cisco.com>
|
|
Force pacing for fast retransmit to avoid bursts of retransmitted
packets.
Change-Id: I2ff42c328899b36322c4de557b1f7d853dba8fe2
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Idbc7b61393c6d0e3b8ea950397a89d21b1cf3a42
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ib138b6e2eac47acc16e81bc88358ae7947420134
Signed-off-by: Eyal Bari <ebari@cisco.com>
|
|
If sessions cannot be handled during the current dispatch loop
iteration, ensure that they are first to be handled in the next.
Change-Id: Ifc6215900f8cfd530d4886b58641189f0ccf9bb7
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I66ca0ddea872948507d078e405eb90f9f3a0e897
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Iaecf8c060e1337d8c362ad9a9be2bb9701664397
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: Ia68db22b917e9af1394c00e5a6b3df134bfd1568
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: Ibef46e068cd72415af28920b0146adf48105bf68
Signed-off-by: Klement Sekera <ksekera@cisco.com>
|
|
Change-Id: I93c6b7bccd1a1ab71625ae29c99c974581186c4d
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: I2eafac4ce810fe53454b729d81161ec80d036db7
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: I7531a64d7072d85514ca579827b6ea0e9cef6f08
Signed-off-by: Vijayabhaskar Katamreddy <vkatamre@cisco.com>
|
|
Change-Id: I4921054b4e42bac3442d399681b21b613ce8b681
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
when the last interface is removed from l2 in the bonding group, we should
invoke ethernet_set_rx_direct to allow ip packets to go directly to
ip4-input.
Change-Id: I43b3cd64e2c119762edd0c295bb9348732adab45
Signed-off-by: Steven <sluong@cisco.com>
|
|
Change-Id: Ie4c5cfc4c97acb321a46b4df589dc44de1b616ba
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I77af4f3a7e826ea5c1a23ee8b348faefe9f2facc
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
length in ip4 header could be 0, when the length edit is fixed
this happens if length is not specified or size is specified as, f.g., 100-100
As a result, tcp and icmp would get a negative value for checksum calculation
Change-Id: I55fa1f5e95717ea4149cb3b8c9b73caf88ae7f98
Signed-off-by: Kingwel Xie <kingwel.xie@ericsson.com>
|
|
stream->max_packet_bytes
This length is the payload length, and will impact the ip length
when ip length is not specified. iplen = header_len + payload_len
SO, better to make it comply with max_packet_bytes
Change-Id: I8b0f7485e29fcaccae656f2d03b3b5e614300fb8
Signed-off-by: Kingwel Xie <kingwel.xie@ericsson.com>
|
|
f.g., gtpu4/6
Change-Id: I8bb1dc5fd2fba89ff17ec069a9816bafb9684190
Signed-off-by: Kingwel Xie <kingwel.xie@ericsson.com>
|
|
clang will emit a warning when the wrong enum type is passed to a
function whose arguments are an enum type. free bug finding...
Change-Id: I62215d8ef22c7527a31272e31f5d190e4e762e53
Signed-off-by: Neale Ranns <nranns@cisco.com>
|
|
Change-Id: I6e43953a6ad1bd672e69d8377d18bd9614b469d8
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Change-Id: I3a47c71ad3e35df47d11fed6db95019a45f3015f
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Avoid possible null pointer dereference
Change-Id: If8023edb43aaf037234f4a7b5f191cb23b09c74d
Signed-off-by: Paul Vinciguerra <pvinci@vinciconsulting.com>
|
|
Change-Id: Ie7827b6a31968a355687d27325c0f30cab1bc890
Signed-off-by: Paul Vinciguerra <pvinci@vinciconsulting.com>
|
|
Change-Id: I91c9d040fc9b9b63f7109eeaac334c47fb1226cf
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Fixed consistancy between the first classifier match and the
next_table_index classifier matches:
- CLASSIFY_ACTION_SET_METADATA was applied only for the first
classifier, but it was not applied for the other classifiers
- Actions should be performed only for input ACLs
- Payload should point at the IP header for output ACLs
Change-Id: Ifbd7791756320ae3198520c41902f5e99e3d40b4
Signed-off-by: Gregory Thiemonge <gregory.thiemonge@enea.com>
|
|
Change-Id: I010528055b9d0597b087882146496eacc13b7daa
Signed-off-by: Paul Vinciguerra <pvinci@vinciconsulting.com>
|
|
Adds tx pacing infrastructure for transport protocols that want to use
it. Particularly useful for connections with non-negligible rtt and
constrained network throughput as it avoids large tx bursts that lead to
local interface tx or network drops.
By default the pacer is disabled. To enabled it for tcp, add tx-pacing
to tcp's startup conf. We are still slightly inefficient in the handling
of incoming packets in established state so the pacer slightly affect
maximum throughput in low lacency scenarios.
Change-Id: Id445b2ffcd64cce015f75b773f7d722faa0f7ca9
Signed-off-by: Florin Coras <fcoras@cisco.com>
|