Commit Graph

8 Commits

Author SHA1 Message Date
Jerin Jacob
3abcd29f2d update Cavium Inc copyright headers
Replace the incorrect reference to "Cavium Networks", "Cavium Ltd"
company name with correct the "Cavium, Inc" company name in
copyright headers.

Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
2017-07-08 17:43:49 +02:00
Ashwin Sekhar T K
3f98dd87d2 efd: support lookup using NEON intrinsics
* Added file lib/librte_efd/rte_efd_arm64.h to hold arm64
  specific definitions
* Verified the changes with efd_autotest unit test case

Signed-off-by: Ashwin Sekhar T K <ashwin.sekhar@caviumnetworks.com>
2017-07-03 15:27:32 +02:00
Adrien Mazarguil
c67ae2bd09 efd: fix missing include in exported header
This commit addresses the following compilation errors:

 In file included from /tmp/check-includes.sh.8373.c:1:0:
 build/include/rte_efd.h:133:9: error: unknown type name 'uint8_t'
 [...]

Fixes: 56b6ef874f ("efd: new Elastic Flow Distributor library")
Cc: stable@dpdk.org

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
2017-05-01 00:13:15 +02:00
Olivier Matz
feb9f680cd mk: optimize directory dependencies
Before this patch, the management of dependencies between directories
had several issues:

- the generation of .depdirs, done at configuration is slow: it can take
  more than one minute on some slow targets (usually ~10s on a standard
  PC without -j).

- for instance, it is possible to express a dependency like:
  - app/foo depends on lib/librte_foo
  - and lib/librte_foo depends on app/bar
  But this won't work because the directories are traversed with a
  depth-first algorithm, so we have to choose between doing 'app' before
  or after 'lib'.

- the script depdirs-rule.sh is too complex.

- we cannot use "make -d" for debug, because the output of make is used for
  the generation of .depdirs.

This patch moves the DEPDIRS-* variables in the upper Makefile, making
the dependencies much easier to calculate. A DEPDIRS variable is still
used to process library dependencies in LDLIBS.

After this commit, "make config" is almost immediate.

Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Tested-by: Robin Jarry <robin.jarry@6wind.com>
Tested-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
2017-03-27 23:28:43 +02:00
Pablo de Lara
b09efeb9d5 doc: add thread-safety information about EFD library
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: John McNamara <john.mcnamara@intel.com>
2017-02-14 21:48:36 +01:00
Olivier Matz
cb8fac62be efd: fix build by removing dependency to libmath
When we compile the dpdk with:
  CONFIG_RTE_LIBRTE_EFD=y
  CONFIG_RTE_LIBRTE_NFP_PMD=n
  CONFIG_RTE_LIBRTE_THUNDERX_NICVF_PMD=n
  CONFIG_RTE_LIBRTE_SCHED=n
  CONFIG_RTE_LIBRTE_METER=n

The linker gives the following error:
  lib/librte_efd.a(rte_efd.o): In function `rte_efd_create':
  lib/librte_efd/rte_efd.c:560: undefined reference to `log2'
  collect2: error: ld returned 1 exit status

This is because the '-lm' is missing in mk/rte.app.mk.

An alternative, which is proposed by this patch, is to use the compiler
builtin rte_bsf32() to process log2 instead of the libmath log2() that
requires to include math.h and link with -lm.

Fixes: 56b6ef874f ("efd: new Elastic Flow Distributor library")

Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
2017-01-30 10:58:40 +01:00
Pablo de Lara
86d8989688 efd: add AVX2 vector lookup function
Signed-off-by: Byron Marohn <byron.marohn@intel.com>
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
Acked-by: Christian Maciocco <christian.maciocco@intel.com>
2017-01-18 20:53:45 +01:00
Pablo de Lara
56b6ef874f efd: new Elastic Flow Distributor library
Elastic Flow Distributor (EFD) is a distributor library that uses
perfect hashing to determine a target/value for a given incoming flow key.
It has the following advantages:

- First, because it uses perfect hashing, it does not store
  the key itself and hence lookup performance is not dependent
  on the key size.

- Second, the target/value can be any arbitrary value hence
  the system designer and/or operator can better optimize service rates
  and inter-cluster network traffic locating.

- Third, since the storage requirement is much smaller than a hash-based
  flow table (i.e. better fit for CPU cache), EFD can scale to
  millions of flow keys.
  Finally, with current optimized library implementation performance
  is fully scalable with number of CPU cores.

Signed-off-by: Byron Marohn <byron.marohn@intel.com>
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
Acked-by: Christian Maciocco <christian.maciocco@intel.com>
2017-01-18 20:53:28 +01:00