Age | Commit message (Collapse) | Author | Files | Lines |
|
Type: refactor
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I4ee46a6c3c53c58199c275e20702f7fd11b60d9a
|
|
- Allowing application to register custom fifo-tuning-logic.
- Adding an example custom fifo-tuning-logic in hs_app/proxy.
Type: feature
Signed-off-by: Ryujiro Shibuya <ryujiro.shibuya@owmobility.com>
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I2aca14d1f23d5c3c9debb7f4c46aca3a15a8d1b9
|
|
1. segment manager would attempt to balance the usages across
the segments, when it allocate fifos
2. the memory presure level is determined per fifo-segment
3. updated unit test
4. updated cli output for segments
Type: feature
Signed-off-by: Ryujiro Shibuya <ryujiro.shibuya@owmobility.com>
Change-Id: I2923f3e0a43dd919196a0cb2cd55e098fde6cf66
|
|
Type: refactor
Switch from a wrapped byte space to a "continuous" one wherein fifo
chunks are appended to the fifo as more data is enqueued and chunks are
removed as data is dequeued.
The fifo is still subject to a maximum size, i.e., maximum number of
bytes that can be enqueued, so the max number of chunks associated to
the fifo is also constrained.
When enqueueing data, which must fit within the available free space, if
not enough "supporting" chunk memory is available, the fifo asks the
fifo segment for enough chunk memory to ensure that the write can
succeed. To avoid allocating large amounts of small chunks due to small
writes, if possible, the size of the chunks requested is lower capped by
min_alloc.
When dequeuing data, all the chunks that have been completely drained,
i.e., head moved beyond the chunks’ end bytes, are unlinked from the
fifo and returned to the fifo segment. The one exception to this is the
last chunk which is never unlinked.
Change-Id: I98c1dbd9135fb79650365c7e40c29238b96cd4ee
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Type: feature
Change-Id: I7d3017bbb369261d74f51807a226f2c12f45291c
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Type: refactor
Change-Id: I9be652e56cdb48b0aee3253f7ce8d9bed299d824
Signed-off-by: Florin Coras <fcoras@cisco.com>
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Previous implementation of clib_rwlock_t used two spinlocks: one
writer lock, and one to guard the counter for the number of readers.
This implementation uses a single condition variable rw_cnt which
has the following properties:
if a writer has the rwlock, rw_cnt = -1
if the rwlock is free, rw_cnt = 0
otherwise, rw_cnt > 0 and rw_cnt = number of readers
rw_cnt will never be less than -1
Benchmarking:
The results below are the cycle counts from test_rwlock.c, configured so
that for 10000 iterations, 6 reader and 6 writer threads on separate cores
are spawned such that each writer thread increments a global counter
10000 times in each iteration. For Taishan, 4 reader and 4 writer
threads are spawned in each test.
x86 Xeon old rwlock: 12.473e8, 11.655e8, 13.201e8, 11.347e8, 13.182e8
x86 Xeon new rwlock: 5.881e8, 5.796e8, 6.536e8, 5.540e8, 5.890e8
Aarch64 ThX2* old rwlock: 9.263e7, 8.933e7, 9.074e7, 8.979e7, 9.378e7
Aarch64 ThX2* new rwlock: 7.221e7, 8.107e7, 7.515e7, 7.672e7, 7.386e7
A72 old rwlock: 3.268e6, 3.200e6, 3.086e6, 3.176e6, 3.170e6
A72 new rwlock: 1.261e6, 1.288e6, 1.251e6, 1.229e6, 1.234e6
*ThunderX2 used additional gcc options "-march=armv8.1-a+crc+crypto+lse"
Type: refactor
Change-Id: I7c347d3037b36205ab532cbcb52a374c846eb275
Signed-off-by: Jason Zhang <jason.zhang2@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Lijian Zhang <Lijian.Zhang@arm.com>
|
|
Type: refactor
Change-Id: I6933205cfb04bc31cabe6e3b1a8044cace93f84c
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- 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>
|
|
- fix segment size rounding for 4GB segments
- fix initialization of first segment size
- cleanup fifo segment info retrieval
Change-Id: I5ebf20f71ea797087653e7e76fa2e37b2686ec40
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ie519683bb90aae6fb95f2a09e251cded1890ed41
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ie96706b4d8bcb32d2d5f065bc765f95f4e9369e7
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I984f347fb465c0c405cef668d8690457e81788e2
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I50a6bcc127e4b44becc4b694bdd3018ac9bfab5c
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I0ef3115bd29a11538090c582a4eacdbb7cd86d7a
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ib346570daa3e40f4f53100a05e9355ce60d533a4
Signed-off-by: Florin Coras <fcoras@cisco.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>
|
|
Change-Id: I42ef9ee38f93600a0d6f2699b1b2a0a201fcec9c
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I3774f7ffd14be73e9bbbf4cdd6d3b22f24fee730
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- Switches local connects to cut-thru transport
- Removes local sessions as a separate session type
Change-Id: I997c6355d8c8e4f2110678f785b0f5d96bba47f7
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Move local session to separate header and source files. First step to
refactoring local sessions.
Change-Id: I280fdfef20ba8a0977d15c1c8ce030ea2fb72dde
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Move app worker logic to app_worker.c
Change-Id: Ic5e5735b2884f006c064d023f491aa6888114810
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Rename core data structures. This will break compatibility for out of
tree builtin apps.
- stream_session_t to session_t
- server_rx/tx_fifo to rx/tx_fifo
- stream_session.h to session_types.h
- update copyright
Change-Id: I414097c6e28bcbea866fbf13b8773c7db3f49325
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Avoid allocating session and possibly reallocating thread session pool
on builtin session rx.
Change-Id: I70e7c604678b44ce8d22603489e247a2c5faa439
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
* u32/u64/uword mismatches
* pointer-to-int fixes
* printf formatting issues
* issues with incorrect "ULL" and related suffixes
* structure alignment and padding issues
Change-Id: I70b989007758755fe8211c074f651150680f60b4
Signed-off-by: David Johnson <davijoh3@cisco.com>
|
|
In addition to that, a bit of refactoring.
Change-Id: Iea1eabc2167bcdef185ec53bc09bae087c5398e6
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I03884b6cde9d4c38ae13d1994fd8d37d44016ef0
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ied34720ca5a6e6e717eea4e86003e854031b6eab
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Change-Id: I0d42a0c71fea7dd669fb1fe5ded7e6e944245c7d
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I6201a044e70ab6a58db8212960c57edc77c41f96
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
If sessions are marked as blocking, events for other sessions received
while waiting for the blocking sessions, are added to a pending list and
processed later.
Change-Id: Ia6c71006b1c2bcb78af708390da0cd436af397cc
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Allows app workers to listen on the same session endpoint. Incoming
connects are spread across the workers in a round-robin fashion
Change-Id: Ib5f5817230d9abc6127a85cdbdcad70d980c0f7f
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Refactor session layer to support multiple workers per application.
Change-Id: Ie67354688d396449d14bbbb8c56050206e307cd8
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- support eventfd based mq signaling. Based on configuration, vcl
epoll/select can use either condvars or epoll on mq eventfds.
- add vcl support for memfd segments
- vpp explicitly registers cut-through segments with apps/vcl
- if using eventfd, make ldp allow one call to libc_epoll_create. Needed
for the message queue epfd
- update svm_queue_t to allow blocking calls with eventfd signaling.
Change-Id: I064151ac370bbe29bb16c968bf4e3659c8286bea
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I242056bc46ddb671064665916b2687860292dcb2
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I3c58367eec2243fe19b75be78a175c5261863e9e
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ibbba75d069ca1bbf9e5a1b8bd2f405d32021c656
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Select the right segment manager for local sessions established via
global table.
Change-Id: I88ad4bf70d0cae160a0c744950098a954dfbc911
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ibc5b1149f3fbbe3dfe1f069ab69bfed7de660582
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Memfd backed shared memory segments can only be negotiated over sockets.
For such scenarios, the existing redirect mechanism that establishes
cut-through sessions does not work anymore as the two peer application
do not share such a socket.
This patch adds support for local sessions, as opposed to sessions
backed by a transport connection, in a way that is almost transparent to
the two applications by reusing the existing binary api messages.
Moreover, all segment allocations are now entirely done through the
segment manager valloc, so segment overlaps due to independent
allocations previously required for redirects are completely avoided.
The one notable characteristic of local sessions (cut-through from app
perspective) notification messages is that they carry pointers to two
event queues, one for each app peer, instead of one. For
transport-backed sessions one of the queues can be inferred but for
local session they cannot.
Change-Id: Ia443fb63e2d9d8e43490275062a708f039038175
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- use valloc as a 'central' segment baseva manager
- use per segment manager segment pools and use rwlocks to guard them
- add session test that exercises segment creation
- embed segment manager properties into application since they're shared
- fix rw locks
Change-Id: I761164c147275d9e8a926f1eda395e090d231f9a
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: If421bad17b6cfe8e321257c93bb38931e37d7b64
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- update segment manager and session api to work with both flavors of
ssvm segments
- added generic ssvm slave/master init and del functions
- cleanup/refactor tcp_echo
- fixed uses of svm fifo pool as vector
Change-Id: Ieee8b163faa407da6e77e657a2322de213a9d2a0
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- separate client/server code for both memory and socket apis
- separate memory api code from generic vlib api code
- move unix_shared_memory_fifo to svm and rename to svm_fifo_t
- overall declutter
Change-Id: I90cdd98ff74d0787d58825b914b0f1eafcfa4dc2
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I39d21d15677f57e10b69b8842f2cbca277abddf0
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I280fea2610dcfc0b2da84973b9f567daec42f1f6
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: Ie747b490901254e962cf61814491851b891129ee
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Change-Id: I44d5c9df7c49b8d4d5677c6d319033b2da3e6b80
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
Applications are now provided the option to select the namespace they
are to be attached to and the scope of their attachement. Application
namespaces are meant to:
1) constrain the scope of communication through the network by
association with source interfaces and/or fib tables that provide the
source ips to be used and limit the scope of routing
2) provide a namespace local scope to session layer communication, as
opposed to the global scope provided by 1). That is, sessions can be
established without assistance from transport and network layers.
Albeit, zero/local-host ip addresses must still be provided in session
establishment messages due to existing application idiosyncrasies. This
mode of communication uses shared-memory fifos (cut-through sessions)
exclusively.
If applications request no namespace, they are assigned to the default
one, which at its turn uses the default fib. Applications can request
access to both local and global scopes for a namespace. If no scope is
specified, session layer defaults to the global one.
When a sw_if_index is provided for a namespace, zero-ip (INADDR_ANY)
binds are converted to binds to the requested interface.
Change-Id: Ia0f660bbf7eec7f89673f75b4821fc7c3d58e3d1
Signed-off-by: Florin Coras <fcoras@cisco.com>
|
|
- add preallocated segment flag
- don't remove pre-allocated segments except if application detaches
- when preallocating fifos in multiple segments, completely fill
a segment before moving to the next
- detach server application from segment-managers when deleting app
- batch syn/syn-ack/fin (re)transmissions
- loosen up close-wait and time-wait times
Change-Id: I412f53ce601cc83b3acc26aeffd7fa2d52d73b03
Signed-off-by: Florin Coras <fcoras@cisco.com>
|