dpdk.git
2 years agonet/iavf: support quanta size configuration
Wenjun Wu [Fri, 22 Apr 2022 01:43:00 +0000 (09:43 +0800)]
net/iavf: support quanta size configuration

This patch adds quanta size configuration support.
Quanta size should between 256 and 4096, and be a product of 64.

Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/iavf: support queue rate limit configuration
Wenjun Wu [Fri, 22 Apr 2022 01:42:59 +0000 (09:42 +0800)]
net/iavf: support queue rate limit configuration

This patch adds queue rate limit configuration support.
Only max bandwidth is supported.

Signed-off-by: Ting Xu <ting.xu@intel.com>
Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/iavf: remove extra copy step in Rx bulk path
Kathleen Capella [Thu, 24 Mar 2022 22:11:32 +0000 (22:11 +0000)]
net/iavf: remove extra copy step in Rx bulk path

In the Rx bulk path, packets which are taken from the HW ring, are first
copied to the stage data structure and then later copied from the stage
to the rx_pkts array. For the number of packets requested immediately
by the receiving function, this two-step process adds extra overhead
that is not necessary.

Instead, put requested number of packets directly into the rx_pkts array
and only stage excess packets. On N1SDP with 1 core/port, l3fwd saw up
to 4% performance improvement. On x86, no difference in performance was
observed.

Signed-off-by: Kathleen Capella <kathleen.capella@arm.com>
Suggested-by: Dharmik Thakkar <dharmik.thakkar@arm.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/ice: fix raw flow input pattern parsing
Ting Xu [Mon, 18 Apr 2022 06:59:08 +0000 (14:59 +0800)]
net/ice: fix raw flow input pattern parsing

When parsing raw flow pattern in FDIR, the input parameter spec and
mask are used directly and the original value will be changed. It
will cause error if these values are used in other functions. In this
patch, temporary variables are created to store the spec and mask.

Fixes: 25be39cc1760 ("net/ice: enable protocol agnostic flow offloading in FDIR")
Cc: stable@dpdk.org
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Junfeng Guo <junfeng.guo@intel.com>
2 years agonet/ice: refactor parser usage
Qi Zhang [Thu, 7 Apr 2022 16:06:14 +0000 (12:06 -0400)]
net/ice: refactor parser usage

Not necessary to create / destroy a parser instance for every raw packet
rule. A global parser instance will be created in ice_flow_init and be
destroyed in ice_flow_uninit.

Also, ice_dev_udp_tunnel_port_add has been hooked to perform corresponding
parser configure. This also fix the issue that RSS engine can't support
VXLAN inner through raw packet filter.

Fixes: 1b9c68120a1c ("net/ice: enable protocol agnostic flow offloading in RSS")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Xu Ting <ting.xu@intel.com>
2 years agonet/ice: add missing Tx burst mode name
Michael Pfeiffer [Mon, 4 Apr 2022 14:41:44 +0000 (16:41 +0200)]
net/ice: add missing Tx burst mode name

The function ice_xmit_pkts_vec_avx2_offload was left out in the list
of tx functions for ice_tx_burst_mode_get.

Fixes: 52ccdcf2fd41 ("net/ice: add AVX2 offload Tx")
Cc: stable@dpdk.org
Signed-off-by: Michael Pfeiffer <michael.pfeiffer@tu-ilmenau.de>
Suggested-by: Michael Rossberg <michael.rossberg@tu-ilmenau.de>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/i40e: populate error in flow director parser
Mike Pattrick [Tue, 22 Mar 2022 03:19:37 +0000 (23:19 -0400)]
net/i40e: populate error in flow director parser

Errors from i40e_flow_parse_fdir_pattern() can bubble up to
rte_flow_create. If rte_flow_error is not initialized a caller may
dereference error->message. This may be uninitialized memory, leading
to a segemntation fault.

Fixes: 4a072ad43442 ("net/i40e: fix flow director config after flow validate")
Cc: stable@dpdk.org
Signed-off-by: Mike Pattrick <mkp@redhat.com>
Reviewed-by: David Marchand <david.marchand@redhat.com>
2 years agonet/ice: improve performance of Rx timestamp offload
Wenjun Wu [Mon, 28 Feb 2022 07:36:07 +0000 (15:36 +0800)]
net/ice: improve performance of Rx timestamp offload

Previously, each time a burst of packets is received, SW reads HW
register and assembles it and the timestamp from descriptor together to
get the complete 64 bits timestamp.

This patch optimizes the algorithm. The SW only needs to check the
monotonicity of the low 32bits timestamp to avoid crossing borders.
Each time before SW receives a burst of packets, it should check the
time difference between current time and last update time to avoid
the low 32 bits timestamp cycling twice.

The patch proved a 50% ~ 70% single core performance improvement on a
main stream Xeon server, this fix the performance gap for some use cases.

Fixes: f9c561ffbccc ("net/ice: fix performance for Rx timestamp")
Cc: stable@dpdk.org
Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/i40e: remove redundant number of packets check
Feifei Wang [Thu, 3 Mar 2022 01:53:43 +0000 (01:53 +0000)]
net/i40e: remove redundant number of packets check

For i40e_xmit_pkts_vec_xx function, it checks nb_pkts to ensure nb_pkts
does not cross rs_thresh.

However, in i40e_xmit_fixed_burst_vec_xx function, this check will be
performed again. To improve code, delete this redundant check.

Suggested-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Signed-off-by: Feifei Wang <feifei.wang2@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/iavf: fix HW ring scan method selection
Steve Yang [Mon, 14 Mar 2022 09:31:46 +0000 (09:31 +0000)]
net/iavf: fix HW ring scan method selection

When setup Rx queue, the rxdid would be changed if it's
"IAVF_RXDID_LEGACY_0/1", that caused the scan HW ring used the wrong
function 'iavf_rx_scan_hw_ring_flex_rxd()'.

Ignore the rxdid changed when equals "IAVF_RXDID_LEGACY_0/1".

Fixes: 0ed16e01313e ("net/iavf: fix function pointer in multi-process")
Cc: stable@dpdk.org
Signed-off-by: Steve Yang <stevex.yang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
2 years agonet/iavf: replace SMP barrier with thread fence in Rx
Kathleen Capella [Mon, 7 Mar 2022 19:26:44 +0000 (19:26 +0000)]
net/iavf: replace SMP barrier with thread fence in Rx

Replace the SMP barrier with atomic thread fence for iavf hw ring scan
in the bulk Rx path.

This patch introduces a change to the iavf driver that was already added
to the i40e driver [1] as part of the adoption of the use of compiler
atomics.

[1]Commit 8649e2356689 ("net/i40e: replace SMP barrier with thread fence
in Rx")

Signed-off-by: Kathleen Capella <kathleen.capella@arm.com>
Reviewed-by: Joyce Kong <joyce.kong@arm.com>
Reviewed-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agonet/ixgbe: retry misbehaving SFP read
Stephen Douthit [Wed, 23 Mar 2022 20:03:46 +0000 (16:03 -0400)]
net/ixgbe: retry misbehaving SFP read

Some XGS-PON SFPs have been observed ACKing I2C reads and returning
uninitialized garbage while their uC boots.  This can lead to the SFP ID
code marking an otherwise working SFP module as unsupported if a bogus
ID value is read while its internal PHY/microcontroller is still
booting.

Retry the ID read several times looking not just for NAK, but also for a
valid ID field.

Since the device isn't NAKing the transaction, the existing longer retry
code in ixgbe_read_i2c_byte_generic_int() doesn't apply here.

Signed-off-by: Stephen Douthit <stephend@silicom-usa.com>
Signed-off-by: Jeff Daly <jeffd@silicom-usa.com>
Reviewed-by: Haiyue Wang <haiyue.wang@intel.com>
2 years agonet/ice: support 256 queues
Wenjun Wu [Thu, 24 Mar 2022 05:12:32 +0000 (13:12 +0800)]
net/ice: support 256 queues

256 queues can be allowed now. This patch improves the code
to support 256 queues for per PF.

Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agotest/bpf: skip test if libpcap is unavailable
Tyler Retzlaff [Tue, 22 Mar 2022 07:12:35 +0000 (00:12 -0700)]
test/bpf: skip test if libpcap is unavailable

test_bpf_convert is being conditionally registered depending on the
presence of RTE_HAS_LIBPCAP except the UT unconditionally lists it as a
test to run.

When the UT runs test_bpf_convert test-dpdk can't find the registration
and assumes the DPDK_TEST environment variable hasn't been defined
resulting in test-dpdk dropping to interactive mode and subsequently
waiting for the remainder of the UT fast-test timeout period before
reporting the test as having timed out.

* unconditionally register test_bpf_convert,
* if ! RTE_HAS_LIBPCAP provide a stub test_bpf_convert that reports the
  test is skipped similar to that done with the test_bpf test.

Fixes: 2eccf6afbea9 ("bpf: add function to convert classic BPF to DPDK BPF")
Cc: stable@dpdk.org
Signed-off-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
2 years agoeal/windows: set main lcore affinity
Tyler Retzlaff [Thu, 14 Apr 2022 12:43:43 +0000 (05:43 -0700)]
eal/windows: set main lcore affinity

Add missing code to affinitize main_lcore from lcore configuration.

Signed-off-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
Reviewed-by: David Marchand <david.marchand@redhat.com>
2 years agoeal: emit warning for unused trylock return value
Mattias Rönnblom [Mon, 11 Apr 2022 15:15:58 +0000 (17:15 +0200)]
eal: emit warning for unused trylock return value

Mark the trylock family of spinlock functions with
__rte_warn_unused_result.

Signed-off-by: Mattias Rönnblom <mattias.ronnblom@ericsson.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
2 years agoeal: add macro to warn for unused function return values
Mattias Rönnblom [Mon, 11 Apr 2022 15:15:57 +0000 (17:15 +0200)]
eal: add macro to warn for unused function return values

This patch adds a wrapper macro __rte_warn_unused_result for the
warn_unused_result function attribute.

Marking a function __rte_warn_unused_result will make the compiler
emit a warning in case the caller does not use the function's return
value.

Signed-off-by: Mattias Rönnblom <mattias.ronnblom@ericsson.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Morten Brørup <mb@smartsharesystems.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
2 years agoexamples/bond: fix invalid use of trylock
Mattias Rönnblom [Mon, 11 Apr 2022 15:15:59 +0000 (17:15 +0200)]
examples/bond: fix invalid use of trylock

The conditional rte_spinlock_trylock() was used as if it is an
unconditional lock operation in a number of places.

Fixes: cc7e8ae84faa ("examples/bond: add example application for link bonding mode 6")
Cc: stable@dpdk.org
Signed-off-by: Mattias Rönnblom <mattias.ronnblom@ericsson.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Min Hu (Connor) <humin29@huawei.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
2 years agoeal: factorize lcore main loop
David Marchand [Tue, 5 Apr 2022 16:34:47 +0000 (18:34 +0200)]
eal: factorize lcore main loop

All OS implementations provide the same main loop.
Introduce helpers (shared for Linux and FreeBSD) to handle synchronisation
between main and threads and factorize the rest as common code.
Thread id are now logged as string in a common format across OS.

Note:
- this change also fixes Windows EAL: worker threads cpu affinity was
  incorrectly reported in log.

- libabigail flags this change as breaking ABI in clang builds:
  1 function with some indirect sub-type change:

  [C] 'function int rte_eal_remote_launch(int (void*)*, void*, unsigned
      int)' at eal_common_launch.c:35:1 has some indirect sub-type
      changes:
    parameter 1 of type 'int (void*)*' changed:
      in pointed to type 'function type int (void*)' at rte_launch.h:31:1:
        entity changed from 'function type int (void*)' to 'typedef
          lcore_function_t' at rte_launch.h:31:1
        type size hasn't changed

  This is being investigated on libabigail side.
  For now, we don't have much choice but to waive reports on this symbol.

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
2 years agoeal: cleanup lcore ID hand-over
David Marchand [Tue, 5 Apr 2022 16:34:46 +0000 (18:34 +0200)]
eal: cleanup lcore ID hand-over

So far, a worker thread has been using its thread_id to discover which
lcore has been assigned to it.

On the other hand, as noted by Tyler, the pthread API does not strictly
guarantee that a new thread won't start running eal_thread_loop before
pthread_create writes to &lcore_config[xx].thread_id.

Though all OS implementations supported in DPDK (recently) ensure this
property, it is more robust to have the main thread directly pass
the worker thread lcore.

Signed-off-by: David Marchand <david.marchand@redhat.com>
2 years agomaintainers: update email address
Ferruh Yigit [Mon, 11 Apr 2022 18:05:07 +0000 (19:05 +0100)]
maintainers: update email address

Use new email address.

Signed-off-by: Ferruh Yigit <ferruh.yigit@xilinx.com>
2 years agogpu/cuda: add more NVIDIA devices
Elena Agostini [Wed, 13 Apr 2022 17:38:48 +0000 (17:38 +0000)]
gpu/cuda: add more NVIDIA devices

Add more NVIDIA GPU devices to the gpu/cuda driver list
compatible with GPUDirect RDMA.

Signed-off-by: Elena Agostini <eagostini@nvidia.com>
2 years agonet/dpaa2: fix dpdmux default interface
Tianli Lai [Tue, 29 Mar 2022 22:01:07 +0000 (06:01 +0800)]
net/dpaa2: fix dpdmux default interface

if dpdmux objects created by restool tools with
the argument "--default-if=<if-id-number>", this
function would change it to 1

Fixes: 1def64c2d79e ("net/dpaa2: add dpdmux initialization and configuration")
Cc: stable@dpdk.org
Signed-off-by: Tianli Lai <laitianli@tom.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
2 years agonet/sfc: drop futile null check
Ivan Malov [Sun, 20 Mar 2022 10:11:00 +0000 (13:11 +0300)]
net/sfc: drop futile null check

The NULL pointer check is clearly unneeded.

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
2 years agoeal/windows: add missing C++ include guards
David Marchand [Tue, 5 Apr 2022 13:48:58 +0000 (15:48 +0200)]
eal/windows: add missing C++ include guards

Add missing 'extern "C"' to file.

Fixes: 1db72630da0c ("eal/windows: do not expose private facilities")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
2 years agoci: add Fedora 35 container in GHA
David Marchand [Mon, 4 Apr 2022 14:24:39 +0000 (16:24 +0200)]
ci: add Fedora 35 container in GHA

Build DPDK with Fedora 35 containers.

GHA container support does not allow caching images and docker hub
seems to limit image pulls.
On the other hand, the Fedora project hub does not seem to limit them,
so prefer this hub.
Nevertheless, let's try to be good citizens and cache (once a day) a
prepared image for subsequent builds.
This preparation is done in a first prepare-container-images job.
The rpm-container-builds job then depends on it with a 'needs:' tag.

Differences with builds in Ubuntu GHA vm images:
- tasks are run as root in containers, no need for sudo,
- compiler must be explicitly installed,
- GHA artifacts can't contain a ':' in their name, and must be filtered,
- environment variables are not inherited and must be passed explicitly,

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Aaron Conole <aconole@redhat.com>
2 years agogpu/cuda: add NVIDIA A30X identifier for DPU
Elena Agostini [Mon, 4 Apr 2022 14:43:52 +0000 (14:43 +0000)]
gpu/cuda: add NVIDIA A30X identifier for DPU

A30X GPU code for DPU.

Signed-off-by: Elena Agostini <eagostini@nvidia.com>
2 years agoeal/windows: fix data race when creating threads
Tyler Retzlaff [Thu, 10 Mar 2022 07:35:08 +0000 (23:35 -0800)]
eal/windows: fix data race when creating threads

eal_thread_loop() uses lcore_config[i].thread_id,
which is stored upon the return from CreateThread().
Per documentation, eal_thread_loop() can start
before CreateThread() returns and the ID is stored.

Create lcore worker threads suspended and then subsequently resume to
allow &lcore_config[i].thread_id be stored before eal_thread_loop
execution.

Fixes: 53ffd9f080fc ("eal/windows: add minimum viable code")
Cc: stable@dpdk.org
Signed-off-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
2 years agoeal/x86: remove atomic header include loop
Haiyue Wang [Thu, 24 Mar 2022 08:41:43 +0000 (16:41 +0800)]
eal/x86: remove atomic header include loop

Remove the x86 top atomic header include from the architecture related
header file, since this x86 top atomic header file has included them.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
2 years agomaintainers: add sections for some bbdev drivers
Nicolas Chautru [Mon, 21 Mar 2022 18:19:50 +0000 (11:19 -0700)]
maintainers: add sections for some bbdev drivers

These were implicit from DPDK script but adding
separate reference to make it explicit.
Separate sections for API and PMDs

Signed-off-by: Nicolas Chautru <nicolas.chautru@intel.com>
2 years agomaintainers: update for NFP
Heinrich Kuhn [Fri, 25 Mar 2022 12:01:33 +0000 (14:01 +0200)]
maintainers: update for NFP

Niklas has been appointed the new maintainer for the NFP PMD.
Update the MAINTAINERS file to reflect this.

Signed-off-by: Heinrich Kuhn <heinrich.kuhn@corigine.com>
2 years agodevtools: document ABI check suppression rules
David Marchand [Wed, 23 Mar 2022 09:24:00 +0000 (10:24 +0100)]
devtools: document ABI check suppression rules

Suppression rules are being added during the life of an ABI and cleaned
when bumping the major version.
Sort and document those rules to avoid pruning rules that should be kept.

Signed-off-by: David Marchand <david.marchand@redhat.com>
2 years agoversion: 22.07-rc0
David Marchand [Fri, 18 Mar 2022 14:35:08 +0000 (15:35 +0100)]
version: 22.07-rc0

Start a new release cycle with empty release notes.
Bump version and ABI minor.

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Aaron Conole <aconole@redhat.com>
2 years agoversion: 22.03.0
Thomas Monjalon [Thu, 17 Mar 2022 09:05:06 +0000 (10:05 +0100)]
version: 22.03.0

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
2 years agodoc: update release notes for 22.03
John McNamara [Wed, 16 Mar 2022 17:33:37 +0000 (17:33 +0000)]
doc: update release notes for 22.03

Fix grammar, spelling and formatting of DPDK 22.03 release notes.

Signed-off-by: John McNamara <john.mcnamara@intel.com>
2 years agodoc: add PMD power management errata with RTM and GCC 9
David Hunt [Wed, 9 Mar 2022 13:22:21 +0000 (13:22 +0000)]
doc: add PMD power management errata with RTM and GCC 9

An errata exists where users may see reduced power savings when using
PMD Power Management. This issue occurs when compiling DPDK applications
with GCC-9 on platforms with TSX enabled. In rte_power_monitor_multi(),
the function may return without successfully starting the RTM
transaction (the _xbegin() fails).

Signed-off-by: David Hunt <david.hunt@intel.com>
2 years agoraw/ifpga: fix build with optimization
Wei Huang [Wed, 16 Mar 2022 07:26:30 +0000 (03:26 -0400)]
raw/ifpga: fix build with optimization

Compile failed with cflag optimization=1 on Ubuntu20.04 with GCC10.3,
it reported vendor_id and dev_id may be used uninitialized in function
ifpga_rawdev_fill_info().
Actually it's not the truth, the variables are initialized in function
ifpga_get_dev_vendor_id(). To avoid such compile error, the variables
are initialized when they are defined.

Fixes: 9c006c45d0c5 ("raw/ifpga: scan PCIe BDF device tree")
Cc: stable@dpdk.org
Signed-off-by: Wei Huang <wei.huang@intel.com>
Acked-by: Tianfei Zhang <tianfei.zhang@intel.com>
Acked-by: Rosen Xu <rosen.xu@intel.com>
2 years agodoc: fix telemetry example in cryptodev guide
Sean Morrissey [Wed, 9 Mar 2022 09:33:10 +0000 (09:33 +0000)]
doc: fix telemetry example in cryptodev guide

Blank line added to the final telemetry example for the
cryptography device library as to fix the example
rendering.

Fixes: 1c559ee8469 ("cryptodev: add telemetry endpoint for capabilities")
Cc: stable@dpdk.org
Signed-off-by: Sean Morrissey <sean.morrissey@intel.com>
Acked-by: Ciara Power <ciara.power@intel.com>
2 years agodoc: add arm64 cross file example
Juraj Linkeš [Tue, 25 Jan 2022 13:20:02 +0000 (14:20 +0100)]
doc: add arm64 cross file example

The docs mention modifications and additions to the cross file,
but there is no demonstration of how those should look like.

Signed-off-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Acked-by: Ruifeng Wang <ruifeng.wang@arm.com>
2 years agodoc: update arm64 cross build of numactl
Juraj Linkeš [Tue, 25 Jan 2022 13:20:01 +0000 (14:20 +0100)]
doc: update arm64 cross build of numactl

Numactl cross compilation doesn't work with clang, remove it
and fix the GCC cross compiler executable name.

Signed-off-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Acked-by: Ruifeng Wang <ruifeng.wang@arm.com>
2 years agodoc: add arm64 toolchain names update instructions
Juraj Linkeš [Tue, 25 Jan 2022 13:20:00 +0000 (14:20 +0100)]
doc: add arm64 toolchain names update instructions

The newer versions have an extra -none- in the name.

Signed-off-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Acked-by: Ruifeng Wang <ruifeng.wang@arm.com>
2 years agodoc: add CFLAGS/LDFLAGS alternatives for arm64 cross build
Juraj Linkeš [Tue, 25 Jan 2022 13:19:59 +0000 (14:19 +0100)]
doc: add CFLAGS/LDFLAGS alternatives for arm64 cross build

Remove CFLAGS and LDFLAGS since Meson doesn't support them well enough.
Add Meson alternatives: -Dc_args and -Dc_link_args on the command line
and in cross files.

Signed-off-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Acked-by: Ruifeng Wang <ruifeng.wang@arm.com>
2 years agodoc: simplify Linux paramter rcu_nocbs in PVP benchmark
Tudor Brindus [Tue, 8 Feb 2022 03:50:18 +0000 (22:50 -0500)]
doc: simplify Linux paramter rcu_nocbs in PVP benchmark

From Documentation/admin-guide/kernel-parameters.txt, specifically the
last sentence:

 nohz_full=      [KNL,BOOT,SMP,ISOL]
        The argument is a cpu list, as described above.
        In kernels built with CONFIG_NO_HZ_FULL=y, set
        the specified list of CPUs whose tick will be stopped
        whenever possible. The boot CPU will be forced outside
        the range to maintain the timekeeping.  Any CPUs
        in this list will have their RCU callbacks offloaded,
        just as if they had also been called out in the
        rcu_nocbs= boot parameter.

The kernel or-s the nohz_full cpumask into the rcu_nocbs cpumask at
startup, and uses that.

Signed-off-by: Tudor Brindus <me@tbrindus.ca>
2 years agodoc: add hyperlinks to driver guides in Linux guide
Bruce Richardson [Tue, 15 Mar 2022 22:05:04 +0000 (22:05 +0000)]
doc: add hyperlinks to driver guides in Linux guide

The document roadmap section was missing any mention of the individual
drivers guides which are important for users. Add them to list.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: add hyperlinks to other docs from Linux guide
Bruce Richardson [Tue, 15 Mar 2022 22:05:03 +0000 (22:05 +0000)]
doc: add hyperlinks to other docs from Linux guide

The document roadmap section called out the titles of other documents,
but these are better as hyperlinks.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: add hyperlinks to driver guides in FreeBSD guide
Bruce Richardson [Tue, 15 Mar 2022 22:05:02 +0000 (22:05 +0000)]
doc: add hyperlinks to driver guides in FreeBSD guide

The document roadmap section was missing any mention of the individual
drivers guides which are important for users. Add them to list.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: add hyperlinks to other docs from FreeBSD guide
Bruce Richardson [Tue, 15 Mar 2022 22:05:01 +0000 (22:05 +0000)]
doc: add hyperlinks to other docs from FreeBSD guide

The document roadmap section called out the titles of other documents,
but these are better as hyperlinks.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: change informational warnings to notes in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:51 +0000 (13:45 +0000)]
doc: change informational warnings to notes in Linux guide

There are two warnings in the VFIO section about limitations of VFIO and
limitations on who can bind/unbind devices. Since these don't actually
describe any unsafe conditions, and are more informational, we can
change these to notes. This also helps emphasise the other warnings in
the documents which flag genuine security concerns.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: consolidate VFIO content on Linux drivers page
Bruce Richardson [Wed, 16 Mar 2022 13:45:50 +0000 (13:45 +0000)]
doc: consolidate VFIO content on Linux drivers page

Rather than having separate sections for VFIO and VFIO no-iommu mode, as
well as a separate section further down the document on troubleshooting
VFIO, we can consolidate all these as subsections into a primary VFIO
section. This section starts with the basics of VFIO use, then covers
no-iommu mode, before moving on to the more advanced topics such as
creating VFs and ending with the troubleshooting subsection.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: move UIO section to the end of Linux drivers page
Bruce Richardson [Wed, 16 Mar 2022 13:45:49 +0000 (13:45 +0000)]
doc: move UIO section to the end of Linux drivers page

To further de-emphasise UIO over the alternatives, we can move the UIO
section of the drivers page to the end of the document, giving more
prominence to VFIO and bifurcated drivers.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: split VFIO section in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:48 +0000 (13:45 +0000)]
doc: split VFIO section in Linux guide

The VFIO section of the page about Linux drivers was rather long and
unstructured. This can be improved by splitting it up into subsections,
to cover the specifics of memory limits and creating VFs. When moving
the various text notes into the relevant subsections, we can drop the
note about kernels earlier than 3.6, since DPDK no longer supports
kernels that old.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: emphasise VFIO over UIO-based modules in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:47 +0000 (13:45 +0000)]
doc: emphasise VFIO over UIO-based modules in Linux guide

VFIO is to be strongly preferred over UIO-based modules, so update our
text and examples to only refer to VFIO, giving an initial reference at
the start to UIO as a fallback option.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: move device binding up in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:46 +0000 (13:45 +0000)]
doc: move device binding up in Linux guide

While the details of VFIO and UIO may be of interest to some, most users
of the doc are likely primarily interested in how to bind their devices
to the kernel driver and then move on to running the app. Therefore, the
most important part of the "Linux Drivers" section of the GSG is the
subsection on "Binding and Unbinding", so put that first.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: make UIO safety warning more visible in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:45 +0000 (13:45 +0000)]
doc: make UIO safety warning more visible in Linux guide

The GSG has a note warning that use of UIO is inherently unsafe due to
lack of IOMMU protection. However, this was only flagged as a "NOTE",
meaning it could easily be missed. Changing the rst tag from "note" to
"warning" and moving it to the top of the UIO subsection makes this a
lot more visible to users.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: fix missing note on UIO module in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:44 +0000 (13:45 +0000)]
doc: fix missing note on UIO module in Linux guide

The docs on binding drivers was updated as part of the removal of the
igb_uio module from the main DPDK repo. As part of that update, a note
about uio_pci_generic requiring legacy interrupts was removed, but
should have been kept.

Fixes: 56bb5841fd06 ("kernel/linux: remove igb_uio")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: replace characters for (R) symbol in Linux guide
Bruce Richardson [Wed, 16 Mar 2022 13:45:43 +0000 (13:45 +0000)]
doc: replace characters for (R) symbol in Linux guide

Some IDEs, such as eclipse, complained on save about the use of special
characters in the (R) symbol in linux GSG doc. We can replace those with
the equivalent "|reg|" text, and including isonum.txt.

Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: remove IOMMU pass-through from Linux guide
Bruce Richardson [Thu, 10 Mar 2022 12:38:43 +0000 (12:38 +0000)]
doc: remove IOMMU pass-through from Linux guide

The "Linux Drivers" section of the GSG already notes that, for use of
UIO, the IOMMU must be disabled or put into pass-through mode.
Therefore, there is no need to duplicate this information in the
"additional functionality" section.  Also the kernel configuration
options documented in the section are enabled as standard on all common
distro kernels, so the information should not be needed in a GSG doc.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: drop reference to KNI from Linux guide
Bruce Richardson [Thu, 10 Mar 2022 12:38:42 +0000 (12:38 +0000)]
doc: drop reference to KNI from Linux guide

The KNI library is disabled by default in DPDK and is already documented
in the programmers guide and also in the sample application guide. There
are also in-kernel alternatives to it.  Therefore, we can drop the
(already fairly minimal) reference to it from the Linux GSG.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: shorten details on HPET use
Bruce Richardson [Thu, 10 Mar 2022 12:38:41 +0000 (12:38 +0000)]
doc: shorten details on HPET use

As best we can tell, the HPET timers are not commonly used, so there is
little need to give extensive detail and commentry on them in the Linux
GSG. As such, we can reduce the GSG section to just a single subsection
and also move it down the page below items which are likely of greater
importance.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: expand list of directories in Linux guide
Bruce Richardson [Thu, 10 Mar 2022 12:38:40 +0000 (12:38 +0000)]
doc: expand list of directories in Linux guide

Not all directories were given in the GSG document, but many of those
omitted would be of interest to users, e.g. "doc", "license" and
"usertools" directories. Adding these leaves only "devtools", and
"kernel" as the only undocumented directories, so add them in too for
completeness.

When updating the section, add "including" to the line leading up to the
directory list, indicating that, while the list is currently complete,
it is not guaranteed to always be.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: merge Linux requirements for applications build
Bruce Richardson [Thu, 10 Mar 2022 12:38:39 +0000 (12:38 +0000)]
doc: merge Linux requirements for applications build

When building end-applications linked with DPDK, the only additional
tool needed is pkg-config/pkgconf. However, the standard development
tools meta-packages on most distro's include this as standard, meaning
it does not really require its own section. The one outlier in the
existing text is "alpine" where it is not present when using "libc-dev"
target. However, changing "gcc" and "libc-dev" to "alpine-sdk"
metapackage aligns alpine with the other distros in this regard.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: remove duplicated note about UEFI secure boot
Bruce Richardson [Thu, 10 Mar 2022 12:38:38 +0000 (12:38 +0000)]
doc: remove duplicated note about UEFI secure boot

A note about secure boot not allowing UIO is present in both the system
requirements section and the driver binding section. This fits better in
the driver binding section, so the copy in system requirements can be
removed. The document in general now also emphasises VFIO over UIO more
than when this note was first added, reducing the need for this warning
to be repeated.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: drop note about old chipset from Linux guide
Bruce Richardson [Thu, 10 Mar 2022 12:38:37 +0000 (12:38 +0000)]
doc: drop note about old chipset from Linux guide

The chipsets referenced in the note[1] were all launched in 2012 and are
now discontinued, so we can drop the note about them at this stage.

[1] https://ark.intel.com/content/www/us/en/ark/products/codename/44946/products-formerly-cave-creek.html

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: update build section of FreeBSD guide
Bruce Richardson [Fri, 11 Mar 2022 20:07:52 +0000 (20:07 +0000)]
doc: update build section of FreeBSD guide

Some minor updates for the section on builing DPDK in the GSG:

* update Python 3.7 package name to the 3.8 version
* note that the pyelftools needs to be tied to the python version
* drop reference to jansson library for legacy telemetry
* replace special characters for (R)

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agodoc: add tested platforms with Mellanox NICs
Raslan Darawsheh [Tue, 15 Mar 2022 09:34:28 +0000 (11:34 +0200)]
doc: add tested platforms with Mellanox NICs

Add tested platforms with Mellanox NICs to the 22.03 release notes.

Signed-off-by: Raslan Darawsheh <rasland@nvidia.com>
2 years agodoc: add tested Intel platforms with Intel NICs
Lingli Chen [Thu, 10 Mar 2022 09:46:19 +0000 (17:46 +0800)]
doc: add tested Intel platforms with Intel NICs

Add tested Intel platforms with Intel NICs to v22.03 release note.

Signed-off-by: Lingli Chen <linglix.chen@intel.com>
Acked-by: Xueqin Lin <xueqin.lin@intel.com>
2 years agonet/cnxk: fix build with optimization
Rakesh Kudurumalla [Fri, 4 Mar 2022 14:23:37 +0000 (19:53 +0530)]
net/cnxk: fix build with optimization

Fix the following build error seen with --optimization=1 and
GCC 10.3.0.

drivers/net/cnxk/cnxk_ethdev_mtr.c: In function
‘cnxk_nix_mtr_policy_validate’:
lib/ethdev/rte_mtr_driver.h:188:10: error: ‘str’ may be used
uninitialized in this function [-Werror=maybe-uninitialized]

drivers/net/cnxk/cn10k_rx.h:149:2 error: ‘frag_ptr’ may be
used uninitialized in this function

Bugzilla ID: 939
Fixes: b526599020ef ("net/cnxk: fix build with GCC 12")
Cc: stable@dpdk.org
Reported-by: Ferruh Yigit <ferruh.yigit@intel.com>
Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com>
Tested-by: Daxue Gao <daxuex.gao@intel.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
2 years agonet/mlx5: fix CPU socket ID for Rx queue creation
Thinh Tran [Wed, 9 Mar 2022 19:49:00 +0000 (14:49 -0500)]
net/mlx5: fix CPU socket ID for Rx queue creation

The default CPU socket ID was used while creating the Rx queue and this caused
creation failure in case if hardware was not resided on the default socket.

The patch sets the correct CPU socket ID for the mlx5_rxq_ctrl before
calling the mlx5_rxq_create_devx_rq_resources() which eventually calls
mlx5_devx_rq_create() with correct CPU socket ID.

Fixes: bc5bee028ebc ("net/mlx5: create drop queue using DevX")
Cc: stable@dpdk.org
Signed-off-by: Thinh Tran <thinhtr@linux.vnet.ibm.com>
Reviewed-by: David Christensen <drc@linux.vnet.ibm.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
2 years agonet/mlx5: fix port matching in sample flow rule
Jiawei Wang [Mon, 7 Mar 2022 02:10:42 +0000 (04:10 +0200)]
net/mlx5: fix port matching in sample flow rule

If there are an explicit port match and sample action in the same flow,
mlx5 PMD pushes the explicit port match in the prefix subflow, and
uses the tag item match in the suffix subflow.

The explicit port match was translated into source vport match so
the sample suffix subflow lost this match after flow split.

This patch copies the explicit port match to the sample suffix subflow,
and the latter gets the correct source vport value in the flow matcher.

Fixes: b4c0ddbfcc58 ("net/mlx5: split sample flow into two sub-flows")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
2 years agoversion: 22.03-rc4
Thomas Monjalon [Tue, 15 Mar 2022 01:16:22 +0000 (02:16 +0100)]
version: 22.03-rc4

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
2 years agoeventdev: fix clang C++ include
Bruce Richardson [Fri, 11 Mar 2022 20:05:23 +0000 (20:05 +0000)]
eventdev: fix clang C++ include

When compiling on FreeBSD with clang and include checking enabled,
errors are emitted due to differences in how empty structs/unions are
handled in C and C++, as C++ structs cannot have zero size.

lib/eventdev/rte_eventdev.h:992:2: error:
union has size 0 in C, non-zero size in C++

Since the contents of the union are all themselves of zero size,
the actual union wrapper is unnecessary. We therefore remove it for C++
builds - though keep it for C builds for safety and clarity of
understanding the code. The alignment constraint on the union is
unnecessary in the case where the whole struct is aligned on a 16-byte
boundary, so we add that constraint to the overall structure to ensure
it applies for C++ code as well as C.

Fixes: 1cc44d409271 ("eventdev: introduce event vector capability")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agocryptodev: fix clang C++ include
Bruce Richardson [Fri, 11 Mar 2022 20:05:22 +0000 (20:05 +0000)]
cryptodev: fix clang C++ include

When compiling on FreeBSD with clang and include checking enabled,
errors are emitted due to differences in how empty structs/unions are
handled in C and C++, as C++ structs cannot have zero size.

lib/cryptodev/rte_crypto.h:127:2: error:
union has size 0 in C, non-zero size in C++

Since the contents of the union are all themselves of zero size,
the actual union wrapper is unnecessary. We therefore remove it for C++
builds - though keep it for C builds for safety and clarity of
understanding the code.

Fixes: c0f87eb5252b ("cryptodev: change burst API to be crypto op oriented")
Fixes: d2a4223c4c6d ("cryptodev: do not store pointer to op specific params")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agocompressdev: fix missing space in log macro
Bruce Richardson [Fri, 11 Mar 2022 20:05:21 +0000 (20:05 +0000)]
compressdev: fix missing space in log macro

Building with clang on FreeBSD with chkincs enabled, we get the
following error about a missing space:

lib/compressdev/rte_compressdev_internal.h:25:58: error:
invalid suffix on literal;
C++11 requires a space between literal and identifier [-Wreserved-user-defined-literal]
        rte_log(RTE_LOG_ ## level, compressdev_logtype, "%s(): "fmt "\n", \

Adding in a space between the '"' and 'fmt' removes the error.

Fixes: ed7dd94f7f66 ("compressdev: add basic device management")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agocompressdev: separate out driver-only headers
Bruce Richardson [Fri, 11 Mar 2022 20:05:20 +0000 (20:05 +0000)]
compressdev: separate out driver-only headers

The headers rte_compressdev_pmd.h and rte_compressdev_internal.h are,
as the filenames suggest, headers for building drivers using the
compressdev APIs. As such they should be marked as
"driver_sdk_headers" rather than just "headers" in the meson.build
file.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agoeal/freebsd: add missing C++ include guards
Bruce Richardson [Fri, 11 Mar 2022 20:05:19 +0000 (20:05 +0000)]
eal/freebsd: add missing C++ include guards

Add missing 'extern "C"' to file.

Fixes: 428eb983f5f7 ("eal: add OS specific header file")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
2 years agomaintainers: update for KNI, UIO and pcap
Ferruh Yigit [Wed, 9 Mar 2022 11:25:29 +0000 (11:25 +0000)]
maintainers: update for KNI, UIO and pcap

Won't able to allocate time on these components, resigning from
maintaining them.

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
2 years agoexamples/l3fwd: fix buffer overflow in Tx
Rahul Bhansali [Tue, 11 Jan 2022 12:50:05 +0000 (18:20 +0530)]
examples/l3fwd: fix buffer overflow in Tx

This patch fixes the stack buffer overflow error reported
from AddressSanitizer.
Function send_packetsx4() tries to access out of bound data
from rte_mbuf and fill it into TX buffer even in the case
where no pending packets (len = 0).
Performance impact:- No

ASAN error report:-
==819==ERROR: AddressSanitizer: stack-buffer-overflow on address
0xffffe2c0dcf0 at pc 0x0000005e791c bp 0xffffe2c0d7e0 sp 0xffffe2c0d800
READ of size 8 at 0xffffe2c0dcf0 thread T0
 #0 0x5e7918 in send_packetsx4 ../examples/l3fwd/l3fwd_common.h:251
 #1 0x5e7918 in send_packets_multi ../examples/l3fwd/l3fwd_neon.h:226

Fixes: 96ff445371e0 ("examples/l3fwd: reorganise and optimize LPM code path")
Cc: stable@dpdk.org
Signed-off-by: Rahul Bhansali <rbhansali@marvell.com>
Reviewed-by: Conor Walsh <conor.walsh@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
2 years agoapp/testpmd: fix flow rule with flex input link
Gregory Etelson [Thu, 10 Mar 2022 05:59:53 +0000 (07:59 +0200)]
app/testpmd: fix flow rule with flex input link

Testpmd reads flex item configuration from a JSON file.
Flex item input link description is stored in testpmd
flow item format. For example, `eth type is 0x0800`.
The item description is placed into a general testpmd CLI
flow rule command template and parsed to convert string into
flow item object.

The patch adds the `actions` section to the flow rule template.

Fixes: 59f3a8acbcdb ("app/testpmd: add flex item commands")
Cc: stable@dpdk.org
Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
2 years agoapp/testpmd: fix GTP header parsing in checksum engine
Gregory Etelson [Sun, 13 Mar 2022 09:01:23 +0000 (11:01 +0200)]
app/testpmd: fix GTP header parsing in checksum engine

GTP header can be followed by an optional 32 bits extension.
GTP notifies about the extension presence through the E, S or PN
header bits.

Csum GTP header parser did not check the extension bits value.

The patch updates GTP header length if at-least one of the
extension bits is set.

Fixes: d8e5e69f3a9b ("app/testpmd: add GTP parsing and Tx checksum offload")
Cc: stable@dpdk.org
Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Aman Singh <aman.deep.singh@intel.com>
2 years agoapp/testpmd: fix show RSS RETA on Windows
Adham Masarwah [Sun, 13 Mar 2022 12:57:02 +0000 (14:57 +0200)]
app/testpmd: fix show RSS RETA on Windows

Replaced using strtoul with strtoull when converting to
64-bit mask field.
In Windows strtoul returns 32-bit values which cause an
issue with show RSS RETA.

Fixes: 66c594904ac ("ethdev: support multiple sizes of redirection table")
Cc: stable@dpdk.org
Signed-off-by: Adham Masarwah <adham@nvidia.com>
Acked-by: Aman Singh <aman.deep.singh@intel.com>
2 years agoapp/testpmd: fix L4 checksum in multi-segments
Kevin Liu [Wed, 29 Dec 2021 09:37:02 +0000 (09:37 +0000)]
app/testpmd: fix L4 checksum in multi-segments

Testpmd forwards packets in checksum mode that it needs to calculate
the checksum of each layer's protocol.

In process_inner_cksums, when parsing tunnel packets, inner L4 offset
should be outer_l2_len + outer_l3_len + l2_len + l3_len.

In process_outer_cksums, when parsing tunnel packets, outer L4 offset
should be outer_l2_len + outer_l3_len.

Fixes: e6b9d6411e91 ("app/testpmd: add SW L4 checksum in multi-segments")
Cc: stable@dpdk.org
Signed-off-by: Kevin Liu <kevinx.liu@intel.com>
Acked-by: Yuying Zhang <yuying.zhang@intel.com>
Acked-by: Aman Singh <aman.deep.singh@intel.com>
2 years agoapp/regex: fix number of matches
Gerry Gribbon [Wed, 9 Mar 2022 23:41:52 +0000 (23:41 +0000)]
app/regex: fix number of matches

Depending on number of jobs specified on command line, part of the
data buffer may not get searched, resulting in incorrect number of
matches being reported.

Additional change to ensure the "All Matches" summary outputs the
correct match start locations in the supplied data buffer.

Fixes: de06137cb295 ("app/regex: add RegEx test application")
Cc: stable@dpdk.org
Signed-off-by: Gerry Gribbon <ggribbon@nvidia.com>
Acked-by: Ori Kam <orika@nvidia.com>
2 years agobpf: fix build with some libpcap version on FreeBSD
David Marchand [Thu, 10 Mar 2022 18:30:40 +0000 (19:30 +0100)]
bpf: fix build with some libpcap version on FreeBSD

This is something caught in UNH FreeBSD env.

For some reason [1], the pcap/bpf.h header started to define _BPF_H_.

It happens that the bpf_impl.h internal DPDK header uses this define as
an internal guard.
This triggers a build failure in bpf_convert.c which can't find
RTE_BPF_LOG macro.

Fix the include guard to use the filename and remove _.

1: https://github.com/the-tcpdump-group/libpcap/pull/1074

Fixes: 94972f35a02e ("bpf: add BPF loading and execution framework")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
2 years agocrypto/ipsec_mb: fix GCM requested digest length
Piotr Bronowski [Wed, 9 Mar 2022 18:02:00 +0000 (18:02 +0000)]
crypto/ipsec_mb: fix GCM requested digest length

This patch removes coverity defect CID 375828:
Untrusted value as argument (TAINTED_SCALAR)

Coverity issue: 375828
Fixes: ceb863938708 ("crypto/aesni_gcm: support all truncated digest sizes")
Cc: stable@dpdk.org
Signed-off-by: Piotr Bronowski <piotrx.bronowski@intel.com>
Acked-by: Ciara Power <ciara.power@intel.com>
2 years agocrypto/ipsec_mb: fix GMAC parameters setting
Pablo de Lara [Wed, 9 Mar 2022 10:45:14 +0000 (10:45 +0000)]
crypto/ipsec_mb: fix GMAC parameters setting

AES-GMAC requires plaintext length to be 0 when using AES-GCM,
so only AAD data is used.

Fixes: a501609ea646 ("crypto/ipsec_mb: fix length and offset settings")
Cc: stable@dpdk.org
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: Fan Zhang <roy.fan.zhang@intel.com>
Acked-by: Radu Nicolau <radu.nicolau@intel.com>
Tested-by: Radu Nicolau <radu.nicolau@intel.com>
2 years agonet/af_xdp: fix shared UMEM fill queue reserve
Ciara Loftus [Fri, 11 Mar 2022 13:45:13 +0000 (13:45 +0000)]
net/af_xdp: fix shared UMEM fill queue reserve

Commit 81fe6720f84f ("net/af_xdp: reserve fill queue before socket create")
moves the fill queue reserve logic to before the creation of the socket in
order to suppress kernel logs like:

XSK buffer pool does not provide enough addresses to fill 2047 buffers on
Rx ring 0

However, for queues that share umem, the fill queue reserve must occur
after the socket creation, because the fill queue is not valid until
that point.

This commit uses the umem refcnt value to determine whether the queue is
sharing a umem, and performs the fill queue reservation either before or
after the socket creation, depending on the refcnt value.

The kernel logs will still be seen for the shared umem queues.

Fixes: 81fe6720f84f ("net/af_xdp: reserve fill queue before socket create")

Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
2 years agonet/af_xdp: fix custom program loading with multiple queues
Junxiao Shi [Wed, 9 Mar 2022 21:18:43 +0000 (21:18 +0000)]
net/af_xdp: fix custom program loading with multiple queues

When the PMD is configured to load a custom XDP program, it sets
XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD flag to prevent libbpf from
loading its default XDP program. However, when queue_count is set to
greater than 1, this flag is only set for the first XSK socket but not
for subsequent XSK sockets. This causes XSK socket creation failure.

This commit ensures that XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD flag is
set for all XSK socket creations when custom XDP program is being used.

Fixes: 01fa83c94d7e ("net/af_xdp: workaround custom program loading")
Cc: stable@dpdk.org
Signed-off-by: Junxiao Shi <git@mail1.yoursunny.com>
2 years agonet/qede: fix maximum Rx packet length
Devendra Singh Rawat [Fri, 4 Mar 2022 12:08:33 +0000 (17:38 +0530)]
net/qede: fix maximum Rx packet length

Size of CRC is not added to max_rx_pktlen, due to this bigger sized
packets(size 1480, 1490 1500) are being dropped.
This fix adds RTE_ETHER_CRC_LEN to max_rx_pktlen.

Fixes: 1bb4a528c41f ("ethdev: fix max Rx packet length")
Cc: stable@dpdk.org
Signed-off-by: Devendra Singh Rawat <dsinghrawat@marvell.com>
Signed-off-by: Rasesh Mody <rmody@marvell.com>
2 years agonet/qede: fix Rx bulk
Devendra Singh Rawat [Fri, 4 Mar 2022 12:08:32 +0000 (17:38 +0530)]
net/qede: fix Rx bulk

qede_alloc_rx_bulk_mbufs() was trimming the number of requested
mbufs count to QEDE_MAX_BULK_ALLOC_COUNT.
The Rx callback was ignorant of this trimming and it was always
resetting the number of empty RX BD ring slots to 0.
This resulted in Rx BD ring getting into an inconsistent
state and ultimately the application fails to receive any traffic.

The fix trims the number of requested mbufs count before
making call to qede_alloc_rx_bulk_mbufs().
After qede_alloc_rx_bulk_mbufs() returns successfully, the
number of empty Rx BD ring slots are decremented by the
correct count.

Fixes: 8f2312474529 ("net/qede: fix performance bottleneck in Rx path")
Cc: stable@dpdk.org
Signed-off-by: Devendra Singh Rawat <dsinghrawat@marvell.com>
Signed-off-by: Rasesh Mody <rmody@marvell.com>
2 years agonet/qede: fix Tx completion
Devendra Singh Rawat [Fri, 4 Mar 2022 12:08:31 +0000 (17:38 +0530)]
net/qede: fix Tx completion

Tx completion routine was first incrementing the number of free
slots in Tx ring and then freeing corresponding mbufs in bulk.
In some situations, the number of mbufs freed were less than
number of Tx ring slots freed. This caused Tx ring to get into an
inconsistent state and ultimately application fails to transmit
further traffic.

The fix first updates the Tx ring SW consumer index, then
increments Tx ring free slot number and finally frees the mbuf,
this is done in a single iteration of loop.

Fixes: 2c41740bf19e ("net/qede: get consumer index once")
Fixes: 4996b959cde6 ("net/qede: free packets in bulk")
Cc: stable@dpdk.org
Signed-off-by: Devendra Singh Rawat <dsinghrawat@marvell.com>
Signed-off-by: Rasesh Mody <rmody@marvell.com>
2 years agodoc: mention mlx5 MPRQ incompatibility with external buffers
Alexander Kozyrev [Thu, 10 Mar 2022 05:06:36 +0000 (07:06 +0200)]
doc: mention mlx5 MPRQ incompatibility with external buffers

Multi-Packet Rx queue uses PMD-managed buffers to store packets.
These buffers are externally attached to user mbufs.
This conflicts with the feature that allows using user-managed
externally attached buffers in an application.
Add the corresponding limitation to MLX5 documentation that MPRQ
and external data buffers cannot be used together.

Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
2 years agodoc: fix modify field action description for mlx5
Viacheslav Ovsiienko [Thu, 24 Feb 2022 16:01:36 +0000 (18:01 +0200)]
doc: fix modify field action description for mlx5

This patch adds mlx5 specifics description about
handling the Ethernet type by modify field action
for VLAN-ed traffic.

Fixes: 641dbe4fb053 ("net/mlx5: support modify field flow action")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
2 years agonet/mlx5: fix implicit tag insertion with sample action
Jiawei Wang [Thu, 10 Mar 2022 04:00:10 +0000 (06:00 +0200)]
net/mlx5: fix implicit tag insertion with sample action

A flow rule with sample action was split into two sub-flows,
and the implicit tag action with unique id was added in the prefix
sub-flow, the suffix sub-flow used the tag item to match with that
unique id, and the implicit set tag action was inserted next to
the sample action.

While there's either PUSH VLAN action or ENCAP action preceding the
sample action, implicit set tag action was added after PUSH VLAN or
ENCAP actions, causing flow creation failure due to rdma-core
does not support this action order.

This patch ensures the implicit set tag action is inserted before
either PUSH VLAN or encap action (if any) in the prefix sub-flow.

Fixes: 6a951567c159 ("net/mlx5: support E-Switch mirroring and jump in one flow")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
2 years agonet/mlx5: forbid multiple ASO actions in a single rule
Rongwei Liu [Wed, 9 Mar 2022 12:08:29 +0000 (14:08 +0200)]
net/mlx5: forbid multiple ASO actions in a single rule

For now, only one ASO action is supported in a single flow rule.
Flow rule with more than one ASO action should be rejected in the
validation stage.

Flow rule with action non-shared AGE and COUNT together should be
treated as non-ASO because AGE will fall back to use HW counter,
not ASO hit object.

Group 0 will use HW counter for AGE action even if no COUNT action.

This commit will reject patterns (no matter which group if transfer)
like:
1. group 1 pattern... / end actions age / meter / end
2. group 1 pattern... / end actions conntrack / meter / end
3. group 1 pattern... / end actions age / conntrack... / end

If AGE comes together with COUNT in the above patterns, it's allowed.

Fixes: daed4b6e ("net/mlx5: use aging by counter when counter exists")
Cc: stable@dpdk.org
Signed-off-by: Rongwei Liu <rongweil@nvidia.com>
Acked-by: Xiaoyu Min <jackmin@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
2 years agonet/mlx5: fix sample flow action on trusted device
Jiawei Wang [Wed, 9 Mar 2022 10:19:46 +0000 (12:19 +0200)]
net/mlx5: fix sample flow action on trusted device

A flow rule with sample action will be split into two sub flows,
and a tag action was added implicitly in the sample prefix sub flow,
the reserved metadata regC index was used for this tag action.

The reserved metadata regC was shared with metering action,
for ConnectX-5 trusted device (VF/SF), the reserved metadata regC was
invalid since PF only supported the legacy metering.

This patch adds the checking for the tag index and back to use the
application tag if a failure happened.

Fixes: a9b6ea45bed6 ("net/mlx5: fix tag ID conflict with sample action")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
2 years agonet/mlx5: fix VLAN push action validation
Dariusz Sosnowski [Wed, 9 Mar 2022 09:39:20 +0000 (09:39 +0000)]
net/mlx5: fix VLAN push action validation

Flow domain and direction was validated when OF_PUSH_VLAN action
appears in flow actions. Flow was rejected whenever this action:

- was used in NIC domain, in ingress direction;
- was used in FDB domain, in ingress direction, on ConnectX-5.

This validation logic rejected a valid case when the OF_PUSH_VLAN
action was used when directing traffic to the hairpin queue,
configured in TX implicit mode.

This patch moves code responsible for OF_PUSH_VLAN validation of
domain and direction from flow_dv_validate_push_vlan() to
flow_dv_validate(). Domain and direction are now validated when either
non-hairpin queue is used or hairpin queue is configured in Tx explicit
mode.

Fixes: 96f85ec489db ("net/mlx5: check VLAN push/pop support")
Cc: stable@dpdk.org
Signed-off-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
2 years agonet/i40e: fix build with 16B descriptor on Arm
Joyce Kong [Fri, 4 Mar 2022 07:35:17 +0000 (07:35 +0000)]
net/i40e: fix build with 16B descriptor on Arm

For NEON vector path, there is an issue when building
with 16B descriptor:
incompatible types when assigning to type ‘uint64x2_t’
from type ‘uint32x4_t’
180 |  descs[3] = vbslq_u32(v_desc3_mask, v_zeros,
vreinterpretq_u32_u64(descs[3]));
    |             ^~~~~~~~~

The fix is to add interpretation between type 'uint64x2_t'
and type 'uint32x4_t'.

Fixes: dba9e3bcd9bb ("net/i40e: add flow mark capability to NEON Rx")

Signed-off-by: Joyce Kong <joyce.kong@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Qi Zhang <qi.z.zhang@intel.com>
2 years agogpu/cuda: mark unused GDRCopy functions parameters
Elena Agostini [Thu, 10 Mar 2022 00:06:50 +0000 (00:06 +0000)]
gpu/cuda: mark unused GDRCopy functions parameters

Without enabling gdrcopy, DPDK build has warnings about
unused gdrcopy functions input parameters.

This patch set them to __rte_unused .

Fixes: 24c7759 ("gpu/cuda: map GPU memory with GDRCopy")

Signed-off-by: Elena Agostini <eagostini@nvidia.com>
2 years agovhost: fix FD leak with inflight messages
David Marchand [Tue, 18 Jan 2022 14:53:30 +0000 (15:53 +0100)]
vhost: fix FD leak with inflight messages

Even if unlikely, a buggy vhost-user master might attach fds to inflight
messages. Add checks like for other types of vhost-user messages.

Fixes: d87f1a1cb7b6 ("vhost: support inflight info sharing")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2 years agovhost: fix queue number check when setting inflight FD
Chenbo Xia [Mon, 14 Feb 2022 08:32:37 +0000 (16:32 +0800)]
vhost: fix queue number check when setting inflight FD

In function vhost_user_set_inflight_fd, queue number in inflight
message is used to access virtqueue. However, queue number could
be larger than VHOST_MAX_VRING and cause write OOB as this number
will be used to write inflight info in virtqueue structure. This
patch checks the queue number to avoid the issue and also make
sure virtqueues are allocated before setting inflight information.

Fixes: ad0a4ae491fe ("vhost: checkout resubmit inflight information")
Cc: stable@dpdk.org
Reported-by: Wenxiang Qian <leonwxqian@gmail.com>
Signed-off-by: Chenbo Xia <chenbo.xia@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>