Update ethernet header during route lookup instead of doing
way later while performing Tx burst. Advantages to doing
is at route lookup is that no additional IP version checks
based on packet data are needed and packet data is already
in cache as route lookup is already consuming that data.
This is also useful for inline protocol offload cases
of v4inv6 or v6inv4 outbound tunnel operations as
packet data will not have any info about what is the tunnel
protocol.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
Store security context pointer in lcore Rx queue config and
get it from there in fast path for better performance.
Currently rte_eth_dev_get_sec_ctx() which is meant to be control
path API is called per packet basis. For every call to that
API, ethdev port status is checked.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
Allow larger burst size of vector event mode instead of restricting
to 32. Also restructure traffic type struct to have num pkts first
so that it is always in first cacheline. Also cache align
traffic type struct. Since MAX_PKT_BURST is not used by
vector event mode worker, define another macro for its burst
size so that poll mode perf is not effected.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
Use HW parsed packet type when ethdev supports necessary protocols.
If packet type is not supported, then register ethdev callbacks
for parse packet in SW. This is better for performance as it
effects fast path.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
Enable Tx IPv4 checksum offload only when Tx inline crypto, lookaside
crypto/protocol or cpu crypto is needed.
For Tx Inline protocol offload, checksum computation
is implicitly taken care by HW.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
Acked-by: Konstantin Ananyev <konstantin.v.ananyev@yandex.ru>
This disable the sve_acle flag for cn10k.
For native build, -Dplatform=cn10k will require to
get sve_acle flag parameter in the build.
Performance impact:-
With l3fwd example, lpm lookup performance increased
by ~21% if Neon is used instead of SVE. Hence, disabled
sve_acle flag for cn10k.
Signed-off-by: Rahul Bhansali <rbhansali@marvell.com>
Reviewed-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Ruifeng Wang <ruifeng.wang@arm.com>
An additional check of control flag sve_acle for
RTE_HAS_SVE_ACLE macro to be part of the build.
If any SoC config doesn't have sve_acle flag parameter
then default it will be considered as true.
Signed-off-by: Rahul Bhansali <rbhansali@marvell.com>
Reviewed-by: Chengwen Feng <fengchengwen@huawei.com>
Reviewed-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Acked-by: Ruifeng Wang <ruifeng.wang@arm.com>
The AltiVec header file is defining "vector", except in C++ build.
The keyword "vector" may conflict easily.
As a rule, it is better to use the alternative keyword "__vector".
The DPDK header file rte_altivec.h takes care of undefining "vector",
so the applications and dependencies are free to define the name "vector".
This is a compatibility breakage for applications which were using
the keyword "vector" for its AltiVec meaning.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Tested-by: Ali Alnubani <alialnu@nvidia.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
Acked-by: Ray Kinsella <mdr@ashroe.eu>
The rte_pcapng_init function doesn't exist, so remove it from the docs.
Also fix minor mistakes in the file.
Signed-off-by: Cian Ferriter <cian.ferriter@intel.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
In pcap_tsc_to_ns(), delta * NSEC_PER_SEC will overflow approx 8
seconds after pcap_init is called when using a TSC with a frequency
of 2.5GHz.
To avoid the overflow, update the saved time and TSC value once
delta >= tsc_hz.
Fixes: 8d23ce8f5e ("pcapng: add new library for writing pcapng files")
Cc: stable@dpdk.org
Signed-off-by: Quentin Armitage <quentin@armitage.org.uk>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Add support for hash functions that compute a signature for an array
of bytes read from a packet header or meta-data. Useful for flow
affinity-based load balancing.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Added the rearm counter to the statistics. Updated the learner table
example to the new learner table timer operation.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Enable the pipeline to use the improved learner table timer operation
through the new "rearm" instruction.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Previously, on lookup hit, the hit key had its timer automatically
rearmed with the same timeout in order to prevent its expiration. Now,
a broader set of actions is available on lookup hit, which has to be
managed explicitly: the key can have its timer rearmed with the same
or with a different timeout, or the key timer can be left unmodified.
The latter option allows the key to expire naturally when the timer
eventually runs out, unless the key is hit again and its timer rearmed
at that point. Needed by the TCP connection tracking state machine.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Add support for packet recirculation. The current packet is flagged
for recirculation using the new "recirculate" instruction; on TX, this
flag causes the packet to execute the full pipeline again as if it was
a new packet, except the packet meta-data is preserved. The new
"recircid" instruction can be used to read the pass number in case the
packet goes several times through the pipeline.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
The packet mirroring is configured through slots and sessions, with
the number of slots and sessions set at init.
The new "mirror" instruction assigns one of the existing sessions to a
specific slot, which results in scheduling a mirror operation for the
current packet to be executed later at the time the packet is either
transmitted or dropped.
Several copies of the same input packet can be mirrored to different
output ports by using multiple slots.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Signed-off-by: Kamalakannan R <kamalakannan.r@intel.com>
Add packet clone operation to the output ports in order to support
packet mirroring.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Signed-off-by: Kamalakannan R <kamalakannan.r@intel.com>
Add support for arguments to the default action of regular and learner
tables at initialization time. Until now, only default actions with no
arguments were accepted in the .spec file.
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Signed-off-by: Yogesh Jangra <yogesh.jangra@intel.com>
Fix the emit instruction for the pathological case of all headers to
be emitted being invalid. In this case, the for loop was essentially
skipped and the last emitted header (or an invalid memory location)
getting corrupted by setting its size to 0 through the assignment to
ho->n_bytes right after the for loop.
Fixes: d60dbdc88a ("pipeline: create inline functions for emit instruction")
Cc: stable@dpdk.org
Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Signed-off-by: Venkata Suresh Kumar P <venkata.suresh.kumar.p@intel.com>
Signed-off-by: Yogesh Jangra <yogesh.jangra@intel.com>
On NUMA systems, default cores (0 and 1) might be on different memory
nodes. Double the amount of memory.
Fixes: 9e6b36c34c ("app/testpmd: reduce memory consumption")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Shell used in documentation generation could not run on Windows.
Rewrite scripts in Python.
New scripts use proper path separators and handle paths with spaces.
Signed-off-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>
API documentation index had spaces between link caption and URL,
which may be unsupported by some Markdown implementations.
That is, "[caption](URL)" is valid but "[caption] (URL)" is not.
The problematic behavior is observed with Doxygen on Windows.
Remove the spaces.
Unfortunately, Markdown syntax is not formally specified.
Fixes: 9bf486e606 ("doc: generate HTML for API with doxygen")
Signed-off-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
CSS for API documentation was customized by a shell script
modifying the file that Doxygen produces.
This way CSS code is kept in a script and an extra build step is added.
Move custom style to a plain CSS file.
Use Doxygen capability to attach this extra stylesheet.
Signed-off-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Enable printing of the outer VLAN if flags indicate it is present.
Cc: stable@dpdk.org
Signed-off-by: Ben Magistro <koncept1@gmail.com>
Reviewed-by: Morten Brørup <mb@smartsharesystems.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
Also mark some conditional functions as const.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Acked-by: Vladimir Medvedkin <vladimir.medvedkin@intel.com>
If a /32 route is entered in the RIB the code to traverse
will not see end of the tree. This is due to trying
to do a negative shift which is an undefined in C.
Fix by checking for max depth as is already done in rib6.
Fixes: 5a5793a5ff ("rib: add RIB library")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Vladimir Medvedkin <vladimir.medvedkin@intel.com>
According to RFC791,the options may appear or not in datagrams.
They must be implemented by all IP modules (host and gateways).
What is optional is their transmission in any particular datagram,
not their implementation. So we have to deal with it during the
fragmenting process.
Add some test data for the IPv4 header optional field fragmenting.
Signed-off-by: Huichao Cai <chcchc88@163.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Added new API for flag to enable or disable TC oversubscription
for best effort traffic class at subport level.
By default TC OV is enabled.
Signed-off-by: Marcin Danilewicz <marcinx.danilewicz@intel.com>
Acked-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
This patch support copy, submit, completed and
completed status functionality of DMA driver.
Signed-off-by: Nipun Gupta <nipun.gupta@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Add additional PMD APIs for DPAA2 QDMA driver for configuring
RBP, Ultra Short format, and Scatter Gather support
Signed-off-by: Nipun Gupta <nipun.gupta@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
This patch support basic DMA operations which includes
device capability and channel setup.
Signed-off-by: Nipun Gupta <nipun.gupta@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
The DPAA2 DMA driver is an implementation of the dmadev APIs,
that provide means to initiate a DMA transaction from CPU.
Earlier this was part of RAW driver, but with DMA drivers
added as separate flavor of drivers, this driver is being
moved to DMA drivers.
Signed-off-by: Nipun Gupta <nipun.gupta@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
With DMA devices supported as a separate flavor of devices,
the DPAA2 QDMA driver is moved in the DMA devices.
This change removes the DPAA2 QDMA driver from raw devices.
Signed-off-by: Nipun Gupta <nipun.gupta@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Within ACL rule IPv6 address can be represented in different ways:
either as 4x4B fields, or as 2x8B fields.
Till now, only first format was supported.
Extend test-acl to support both formats, mainly for testing and
demonstrating purposes.
To control desired behavior '--ipv6' command-line option is extended
to accept an optional argument:
To be more precise:
'--ipv6' - use 4x4B fields format (default behavior)
'--ipv6=4B' - use 4x4B fields format (default behavior)
'--ipv6=8B' - use 2x8B fields format
Also replaced home brewed IPv4/IPv6 address parsing with inet_pton() calls.
Signed-off-by: Konstantin Ananyev <konstantin.v.ananyev@yandex.ru>
In theory ACL library allows fields with 8B long.
Though in practice they usually not used, not tested,
and as was revealed by Ido, this functionality is not working properly.
There are few places inside ACL build code-path that need to be addressed.
Bugzilla ID: 673
Fixes: dc276b5780 ("acl: new library")
Cc: stable@dpdk.org
Reported-by: Ido Goshen <ido@cgstowernetworks.com>
Signed-off-by: Konstantin Ananyev <konstantin.v.ananyev@yandex.ru>
Tested-by: Ido Goshen <ido@cgstowernetworks.com>
The AltiVec header file is defining "vector", except in C++ build.
The keyword "vector" may conflict easily.
As a rule, it is better to use the alternative keyword "__vector",
so we will be able to #undef vector after including AltiVec header.
Later it may become possible to #undef vector in rte_altivec.h
with a compatibility breakage.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: David Christensen <drc@linux.vnet.ibm.com>
Current pmd_perf_autotest() in continuous mode tries
to enqueue MAX_TRAFFIC_BURST completely before starting
the test. Some drivers cannot accept complete
MAX_TRAFFIC_BURST even though rx+tx desc count can fit it.
This patch changes behaviour to stop enqueuing after few
retries.
Fixes: 002ade70e9 ("app/test: measure cycles per packet in Rx/Tx")
Cc: stable@dpdk.org
Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com>
Enable GPU_REGISTERED flag in gpu/cuda driver in the memory list.
If a GPU memory address CPU mapped is freed before being
unmapped, CUDA driver unmaps it before freeing the memory.
Signed-off-by: Elena Agostini <eagostini@nvidia.com>
FreeBSD has updated its CPU macros to align more with the definitions
used on Linux[1]. Unfortunately, while this makes compatibility better
in future, it means we need to have both legacy and newer definition
support. Use a meson check to determine which set of macros are used.
[1] https://cgit.freebsd.org/src/commit/?id=e2650af157bc
Bugzilla ID: 1014
Fixes: c3568ea376 ("eal: restrict control threads to startup CPU affinity")
Fixes: b6be16acfe ("eal: fix control thread affinity with --lcores")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Tested-by: Daxue Gao <daxuex.gao@intel.com>
Forcing inlining in test_ring_enqueue and test_ring_dequeue can cause
the compiled code to grow extensively when compiled with no optimization
(-O0 or -Og). This is default in the meson's debug configuration. This
can collide with compiler bugs and cause issues during linking of unit
tests where the api_type or esize are non-const variables causing
inlining cascade. In perf tests this is not the case in perf-tests as
esize and api_type are const values.
One such case was discovered when porting DPDK to RISC-V. GCC 11.2 (and
no fix still in 12.1) is generating a short relative jump instruction
(J <offset>) for goto and for loops. When loop body grows extensively in
ring test, the target offset goes beyond supported offfset of +/- 1MB
from PC. This is an obvious bug in the GCC as RISC-V has a
two-instruction construct to jump to any absolute address (AUIPC+JALR).
However there is no reason to force inlining as the test code works
perfectly fine without it.
GCC has a bug report for a similar case (with conditionals):
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93062
Fixes: a9fe152363 ("test/ring: add custom element size functional tests")
Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Acked-by: Konstantin Ananyev <konstantin.v.ananyev@yandex.ru>
