- track fifo segment free and chunk freelist memory
- improve fifo alloc. If there are enough chunks to satisfy a fifo
allocation request but not enough free memory, allocate a multi-chunk
fifo
- add apis to preallocate chunks and fifo headers
- more tests
Change-Id: If18dba7ab856272c9f565d36ac36365139793e0b
Signed-off-by: Florin Coras <fcoras@cisco.com>
a bit like QUIC, and the added function can get the endpoint info in TLS,
so that the VCL layer can get those info correctly
Change-Id: Ied7aa3077087c1814499364dfa7654a088ad9910
Signed-off-by: Yu Ping <ping.yu@intel.com>
This allows QUIC & TLS specific logic to be implemented, and meaningfull
IP/port to be returned when connection is overridden.
Change-Id: Id79c59fe4d7b16d36f0e96ad3e281c4026b5fe65
Signed-off-by: Aloys Augustin <aloaugus@cisco.com>
Fifos can use multiple memory chunks for simple read/write operations.
Adding/removing chunks after assignment not yet supported.
Change-Id: I2aceab6aea78059d74e0d3a9993c40d5196d077b
Signed-off-by: Florin Coras <fcoras@cisco.com>
Problems Addressed:
- Contention of cursize by producer and consumer.
- Reduce the no of modulo operations.
Changes:
- Synchronization between producer and consumer changed from cursize
to head and tail indexes
Implications: reduces the usable size of fifo by 1.
- Using weaker memory ordering C++11 atomics to access head and tail
based on producer and consumer role.
- Head and tail indexes are unsigned 32 bit integers. Additions and
subtraction on them are implicit 32 bit Modulo operation.
- Adding weaker memory ordering variants of max_enq, max_deq, is_empty
and is_full Using them appropriately in all places.
Perfomance improvement (iperf3 via Hoststack):
iperf3 Server: Marvell ThunderX2(AArch64) - iperf3 Client: Skylake(x86)
~6%(256 rxd/txd) - ~11%(2048 rxd/txd)
Change-Id: I1d484e000e437430fdd5a819657d1c6b62443018
Signed-off-by: Sirshak Das <sirshak.das@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
To connect a stream, apps should call connect while passing the id of the QUIC
connection in the new transport_opts field in session_endpoint_cfg_t.
Apps are notified of new streams with their accept callback, which is called
each time a peer opens a stream.
Change-Id: I0f82ec344db58008d54641553eddec2973768435
Signed-off-by: Aloys Augustin <aloaugus@cisco.com>
This enables applications to create sessions in their RX callbacks, which can
invalidate the session pointer. This is required for the QUIC protocol
implementation.
Change-Id: I6072c1c368fd9d17a960ec086a788089dd6f54b4
Signed-off-by: Aloys Augustin <aloaugus@cisco.com>
Also:
- force reset if wait close pops in fin-wait-1 with unsent data
- adds more event logging.
Change-Id: I4ddada046214fa71e17514cdec57b3026f84a283
Signed-off-by: Florin Coras <fcoras@cisco.com>
- Avoid doing cc in closing states.
- Rest connections closed with unread data
Change-Id: I97d46b0459f03ea5439eeb0f233b6c17d3e06dfd
Signed-off-by: Florin Coras <fcoras@cisco.com>
Currently supports on single stream exposed through standard internal APIs
Based on libquicly & picotls by h2o
Change-Id: I7bc1ec0e399d1fb02bfd1da91aa7410076d08d14
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
This fixes a bug where packets could be sent but not received when opening an UDP connection.
Change-Id: I0993dd806b277374232d551167970ab13a62dbf9
Signed-off-by: Aloys Augustin <aloaugus@cisco.com>
Ensures that fifo cursize loads cannot be speculated to before the event
unset.
Change-Id: Ia7c20c510d58f26a8e9b82d3982c6d4143a3a4d6
Signed-off-by: Florin Coras <fcoras@cisco.com>
- use flag instead of enqueue_epoch for enqueueing rx events.
- use flag for proxy sessions
Change-Id: Iec3eee55a68d02536ece6329348a3369c7c7412e
Signed-off-by: Florin Coras <fcoras@cisco.com>
current session rule cli can not add session rule of proto is udp.
because vnet_session_rule_add_del function's paramter does not include transport_proto element in args struct,
but session rule type on the basis of input args's transport_proto when create session rule with vnet_session_rule_add_del function,
so just add transport_proto element to vnet_session_rule_add_del function's args that to solve this problem
Change-Id: If1a5942b4a0b006d73376e0cb01b97e84c593493
Signed-off-by: zhanglimao <zhanglimao0017@gmail.com>
Instead of allocating pairs of message queues per cut-thru session and
having the applications map them, this uses vpp as an io event message
switch.
Change-Id: I51db1c7564df479a7d1a3288342394251fd188bb
Signed-off-by: Florin Coras <fcoras@cisco.com>
