Provide a bit more diagnostics information when port start fails.
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Acked-by: Xiaoyun Li <xiaoyun.li@intel.com>
The comment got it wrong. The payload length field
does not include the fixed IPv6 header size.
Fixes: 7eca7f7fd0 ("net: add missing endianness annotations")
Fixes: af75078fec ("first public release")
Cc: stable@dpdk.org
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
This commit introduces support for preferred busy polling
to the AF_XDP PMD. This feature aims to improve single-core
performance for AF_XDP sockets under heavy load.
A new vdev arg is introduced called 'busy_budget' whose default
value is 64. busy_budget is the value supplied to the kernel
with the SO_BUSY_POLL_BUDGET socket option and represents the
busy-polling NAPI budget. To set the budget to a different value
eg. 256:
--vdev=net_af_xdp0,iface=eth0,busy_budget=256
Preferred busy polling is enabled by default provided a kernel with
version >= v5.11 is in use. To disable it, set the budget to zero.
The following settings are also strongly recommended to be used in
conjunction with this feature:
echo 2 | sudo tee /sys/class/net/eth0/napi_defer_hard_irqs
echo 200000 | sudo tee /sys/class/net/eth0/gro_flush_timeout
.. where eth0 is the interface being used by the PMD.
Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
poll() is more expensive and requires more tuning
when used with the upcoming busy polling functionality.
Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Prior to this commit, the maximum batch sizes for zero-copy and
copy-mode rx and copy-mode tx were set to 32. Apart from zero-copy tx,
the user could never rx/tx any more than 32 packets at a time and
without inspecting the code the user wouldn't be aware of this.
This commit removes these upper limits placed on the user and instead
sets an internal batch size equal to the default ring size (2048).
Batches larger than this are still processed, however they are split
into smaller batches similar to how it's done in other drivers. This is
necessary because some arrays used during rx/tx need to be sized at
compile-time.
Allowing a larger batch size allows for fewer batches and thus larger
bulk operations, fewer ring accesses and fewer syscalls which should
yield improved performance.
Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
In Windows probing, the value RTE_PCI_KDRV_NONE was used
instead of RTE_PCI_KDRV_UNKNOWN.
This value covers the mlx case where the kernel driver is in place,
offering a bifurcated mode to the userspace driver.
When the kernel driver is listed as unknown,
there is no special treatment in DPDK probing, contrary to UIO modes.
The value RTE_PCI_KDRV_NIC_UIO (FreeBSD) was re-used
instead of having a new RTE_PCI_KDRV_NET_UIO for Windows NetUIO.
While adding the new value RTE_PCI_KDRV_NET_UIO
(at the end for ABI compatibility),
the enum of kernel driver categories is annotated.
Fixes: b762221ac2 ("bus/pci: support Windows with bifurcated drivers")
Fixes: c76ec01b45 ("bus/pci: support netuio on Windows")
Cc: stable@dpdk.org
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Acked-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Ranjit Menon <ranjit.menon@intel.com>
Some functions were introduced in DPDK 21.05 to query the version parts
(prefix, year, month, minor, suffix, release) at runtime.
Per guidelines, these new public functions must be marked with
__rte_experimental and ABI versioned as EXPERIMENTAL.
Fixes: 5b637a8481 ("eal: fix querying DPDK version at runtime")
Cc: stable@dpdk.org
Suggested-by: David Marchand <david.marchand@redhat.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
The macro RTE_VERSION was broken since updated with function calls.
It is a build-time version number, and must be built with macros.
For a run-time version number, there is the function rte_version().
Fixes: 5b637a8481 ("eal: fix querying DPDK version at runtime")
Cc: stable@dpdk.org
Reported-by: David Marchand <david.marchand@redhat.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: David Marchand <david.marchand@redhat.com>
This patch add support to enable Rx offload for timestamp.
It is required to be enabled for some PMDs e.g. dpaa2.
Signed-off-by: Gagandeep Singh <g.singh@nxp.com>
Signed-off-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Acked-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
The headers rte_power_intrinsics.h and rte_power_pmd_mgmt.h
were missing from the doxygen API index.
Fixes: cda57d9388 ("eal: add power management intrinsics")
Fixes: 682a645438 ("power: add ethdev power management")
Cc: stable@dpdk.org
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
EAL -a and -b options are used to specify which PCI devices are
explicitly allowed or blocked during PCI bus scan. This evaluation
is missing in the Windows implementation of rte_pci_scan.
This patch provides this missing functionality, so that apps can specify
which devices to ignore during PCI bus scan.
Signed-off-by: Khoa To <khot@microsoft.com>
Acked-by: Ranjit Menon <ranjit.menon@intel.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Attaching to an NVMe disk on Windows using SPDK requires the
PCI class ID and device.bus fields. Decode the class ID from the PCI
device info strings if it is present and set device.bus.
Signed-off-by: Nick Connolly <nick.connolly@mayadata.io>
Acked-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Implement rte_pci_map_device() to distinguish between the devices bound
to netuio and NDIS devices.
Only return success for the netuio devices.
Fixes: c76ec01b45 ("bus/pci: support netuio on Windows")
Cc: stable@dpdk.org
Suggested-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Signed-off-by: Pallavi Kadam <pallavi.kadam@intel.com>
Reviewed-by: Ranjit Menon <ranjit.menon@intel.com>
Acked-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Narcisa Vasile <navasile@linux.microsoft.com>
Tested-by: Narcisa Vasile <navasile@linux.microsoft.com>
The hard-coded thread priority for Windows threads in EAL
is REALTIME_PRIORITY_CLASS/THREAD_PRIORITY_TIME_CRITICAL.
This results in issues with DPDK threads causing OS thread starvation
and eventually a bugcheck.
The fix reduce the thread priority to
NORMAL_PRIORITY_CLASS/THREAD_PRIORITY_NORMAL.
Bugzilla ID: 600
Fixes: 53ffd9f080 ("eal/windows: add minimum viable code")
Cc: stable@dpdk.org
Reported-by: Odi Assli <odia@nvidia.com>
Signed-off-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
This patch allows the same set of rte_metrics_tel_* functions to be
exported no matter JANSSON is available or not, by doing following:
1. Leverage dpdk_conf to set configuration flag RTE_HAS_JANSSON
when Jansson dependency is found.
2. In rte_metrics_telemetry.c, leverage RTE_HAS_JANSSON to handle the
case when JANSSON is not available by adding stubs for all the instances.
3. In meson.build, per dpdk/doc/guides/rel_notes/release_20_05.rst,
it is claimed that "Telemetry library is no longer dependent on the
external Jansson library, which allows Telemetry be enabled by default.",
thus make the deps and includes of Telemetry as not conditional anymore.
Signed-off-by: Jie Zhou <jizh@microsoft.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
In Linux by default DPDK log goes to stdout, as well as syslog.
It is possible for an application to change the library output stream
via 'rte_openlog_stream()' API, to set it to stderr, it can be used as:
rte_openlog_stream(stderr);
But still updating the default log output to 'stderr'.
Bugzilla ID: 8
Fixes: af75078fec ("first public release")
Cc: stable@dpdk.org
Reported-by: Alexandre Ferrieux <alexandre.ferrieux@orange.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
The KNI linux module is using a custom target for building, which
doesn't take into account any cross compilation arguments. The arguments
in question are ARCH, CROSS_COMPILE (for gcc, clang) and CC, LD (for
clang). Get those from the cross file and pass them to the custom
target.
The user supplied path may not contain the 'build' directory, such as
when using cross-compiled headers, so only append that in the default
case (when no path is supplied in native builds) and use the unmodified
path from the user otherwise. Also modify the install path accordingly.
Signed-off-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>
For using a DPDK application, such as OVS, which is dynamically linked, the
DPDK version in use should always report the actual version, not the
version used at build time. This incorrect behaviour can be seen by
building OVS against one version of DPDK and running it against a later
one. Using "ovs-vsctl list Open_vSwitch" to query basic info, the
dpdk_version returned will be the build version not the currently running
one - which can be verified using the DPDK telemetry library client.
$ sudo ovs-vsctl list Open_vSwitch | grep dpdk_version
dpdk_version : "DPDK 20.11.0-rc4"
$ echo quit | sudo dpdk-telemetry.py
Connecting to /var/run/dpdk/rte/dpdk_telemetry.v2
{"version": "DPDK 21.02.0-rc2", "pid": 405659, "max_output_len": 16384}
-->
To fix this, we need to convert the rte_version() function, and any other
necessary parts of the rte_version.h, to be actual functions in EAL, not
just inlines/macros. The only complication in doing so is that telemetry
library cannot call rte_version() directly, and instead needs the version
string passed in on init.
Fixes: af75078fec ("first public release")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
The meson.build files in each example directory is simply to support
building the example as part of the main SDK build, and these should not
be installed with the example's source code and makefile. The exclude of
"meson.build" only filters out the top-level examples/meson.build file,
not the file in each subdirectory.
To fix this, we can build up the list of files to exclude based off the
list of all examples. With this change "find examples/ -name meson.build"
returns no hits when run on an installed instance.
Fixes: e5b95003f1 ("examples: fix flattening directory layout on install")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
With I/O BAR, we get PIO (port-mapped I/O) address.
With MMIO (memory-mapped I/O) BAR, we get mapped virtual address.
We distinguish PIO and MMIO by their address range like how kernel does,
i.e, address below 64K is PIO.
ioread/write8/16/32 is provided to access PIO/MMIO.
By the way, for virtio on arch other than x86, BAR flag indicates PIO
but is mapped.
Signed-off-by: Huawei Xie <huawei.xhw@alibaba-inc.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tested-by: Yinan Wang <yinan.wang@intel.com>
Currently virtio PMD assumes legacy device uses PIO bar.
There are three ways to get PIO (port-mapped I/O) address for virtio
legacy device.
1) under igb_uio
- get PIO address from uio/uio# sysfs attribute, for instance:
/sys/bus/pci/devices/0000:00:09.0/uio/uio0/portio/port0/start
2) under uio_pci_generic
- for X86, get PIO address from /proc/ioport
- for other ARCH, get PIO address from standard PCI sysfs attribute,
for instance: /sys/bus/pci/devices/0000:00:09.0/resource
Actually, "port0/start" in igb_uio and "resource" point to exactly the
same thing, i.e, pci_dev->resource[0] in kernel source code.
This patch refactors these messy things, and uses standard PCI sysfs
attribute "resource".
Signed-off-by: Huawei Xie <huawei.xhw@alibaba-inc.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tested-by: Yinan Wang <yinan.wang@intel.com>
This patch fixes incorrect MCAM key preparation when creating
MCAM entry to allow VLAN IDs after vlan filtering is enabled on port.
Fixes: ba1b3b081e ("net/octeontx2: support VLAN offloads")
Cc: stable@dpdk.org
Signed-off-by: Satheesh Paul <psatheesh@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
The issue occurred if mbuf starvation happened
in the middle of segmented packet reception.
In such a situation, after release the segments of
packet being received, code did not advance the
consumer index to the next stride. This caused
the receiving of the wrong segmented packet data.
The possible error scenario:
- we assume segs_n is 4 and we are receiving 4
segments of multi-segment packet.
- we fail to allocate mbuf while receiving the 3rd segment,
and this frees the mbufs of the packet chain we have built.
There are the 1st and 2nd segments in the chain.
- the 1st and the 2nd segments of this stride of Rx queue
are filled up (in elts array) with the new allocated
mbufs and their data are random (the 3rd and 4th
segments still contain the valid data of the packet though).
- on the next iteration of stride processing we get
the wrong two segments of the multi-segment packet.
Hence, we should skip these mbufs in the stride and
we should advance the consumer index on loop exit.
Fixes: 15a756b637 ("net/mlx5: fix possible NULL dereference in Rx path")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Zhu <zhujiawei12@huawei.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
IPv4 fragment_offset mask was required to be 0 no matter what the
spec value was. But zero mask means not caring about fragment_offset
field then both non-frag and frag packets should hit the rule.
But the actual fragment rules should be like the following:
Only non-fragment packets can hit Rule 1:
Rule 1: mask=0x3fff, spec=0
Only fragment packets can hit rule 2:
Rule 2: mask=0x3fff, spec=0x8, last=0x2000
This patch allows the above rules.
Fixes: 42044b69c6 ("net/i40e: support input set selection for FDIR")
Cc: stable@dpdk.org
Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
Below miscellaneous APIs are used to implement OPAE application.
1. rte_pmd_ifpga_get_pci_bus() get PCI bus ifpga driver registered.
2. rte_pmd_ifpga_partial_reconfigure() do partial reconfiguration.
3. rte_pmd_ifpga_cleanup() free software resources allocated by driver.
4. rte_pmd_ifpga_set_rsu_status() set status of rsu process.
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>
There are some information data can be got from FPGA, they are
implemented in below APIs:
1. rte_pmd_ifpga_get_property() get properties of FPGA (include BMC).
2. rte_pmd_ifpga_get_phy_info() get information of PHY connect to FPGA.
3. rte_pmd_ifpga_get_rsu_status() get status of rsu process.
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>
RSU (Remote System Update) depends on secure manager which may be
different on various implementations, so a new secure manager device
is implemented for adapting such difference.
There are five APIs added:
1. rte_pmd_ifpga_get_dev_id() get raw device ID of ifpga device from PCI
address like 'Domain:Bus:Dev.Func'.
2. rte_pmd_ifpga_update_flash() update flash with specific image file.
3. rte_pmd_ifpga_stop_update() abort flash update process.
4. rte_pmd_ifpga_reboot_try() check current ifpga status and change it
to reboot status if it is idle.
5. rte_pmd_ifpga_reload() trigger full reconfiguration of ifpga device.
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>
When Tx queue number is more than Rx queue number, and RSS is
enabled, there'll be packet loss with X722.
The root cause is the lookup table is not configured correctly,
since it uses VF's queue pair number but not Rx queue number.
Fixes: 2da3ba7467 ("net/i40e: fix VF runtime queues RSS config")
Cc: stable@dpdk.org
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Signed-off-by: Hengjian Zhang <hengjianx.zhang@intel.com>
Acked-by: Jeff Guo <jia.guo@intel.com>
Currently, FDIR only support GTPU outer fields in PF. Clean the
redundant GTPU inner info in flow type definition and align with
shared code.
Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Distinguish input_set_mask to inner and outer part. Use
input_set_mask_o for tunnel outer or non-tunnel input set.
input_set_mask_i is used for tunnel inner fields only.
Adjust indentation of ice_pattern_match_item list in switch, ACL, RSS
and FDIR for easy review.
For switch, ACL and RSS, only use
input_set_mask_o and set the input_set_mask_i all none.
Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
For tunnel or non-tunnel packet, the input set is in outer_input_set
and use seg_tun[0]. seg_tun[1] is only used for tunnel inner fields.
This patch make align with input_set inner/outer with seg_tun[] and
simplify it.
Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Distinguish inner/outer input_set. And avoid too many nested
conditionals in each type's parser. input_set_o is used for
tunnel outer fields or non-tunnel fields , input_set_i is only
used for inner fields.
For GTPU, store the outer IP fields in inner part to align with
shared code behavior.
Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
This patch use input_set_o and input_set_i to distinguish inner/outer
input set. input_set_i is only used for inner field.
Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Currently, the macro of input set use 2 bits, one bit for protocol and
inner/outer, another bit for src/dst field. But this could not
distinguish a rule with inner and outer fields for tunnel packet.
Redefine input set macro to make it clear. Only use these two bits for
protocol and field. Ignore the redundant inner/outer info.
ICE_INSET_TUN_* is used by switch module, should be removed after
switch refactor.
Signed-off-by: Zhirun Yan <zhirun.yan@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
When ice_remove_vsi_lkup_fltr is called, by calling
ice_add_to_vsi_fltr_list local copy of vsi filter list
is created. If any issues during creation of vsi filter
list occurs it up for the caller to free already
allocated memory. This patch ensures proper memory
deallocation in these cases.
Fixes: c7dd159311 ("net/ice/base: add virtual switch code")
Cc: stable@dpdk.org
Signed-off-by: Robert Malz <robertx.malz@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
One of the structs being used for ACL counter rules was allocated on
the stack and left uninitialized. Rather than depending on
undefined behavior around the .amount member during rule removal,
just leave a comment and initialize the struct to zero, as this is a
slow path call anyway. This bug could have caused silent failures
during counter removal.
Fixes: f3202a097f ("net/ice/base: add ACL module")
Cc: stable@dpdk.org
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
The entry for PTYPE 2 in the ice_ptype_lkup table incorrectly states
that this is an L2 packet with no payload. According to the datasheet,
this PTYPE is actually unused and reserved.
Modify the lookup entry to indicate this is an unused entry that is
reserved.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
The entry for PTYPE 90 indicates that the payload is layer 3. This does
not match the specification in the datasheet which indicates the packet
is a MAC, IPv6, UDP packet, with a payload in layer 4.
Fix the lookup table to match the data sheet.
Fixes: 64e9587d56 ("net/ice/base: add structures for Rx/Tx queues")
Cc: stable@dpdk.org
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Support IPV4_GTPU with inner IPV6/UDP/TCP for FDIR.
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Enabled support for advanced switch filter to satisfy match criteria
such as: GTP tunnel + Inner IPv4[6]
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Store the link_cfg_err byte in order to determine whether an unsupported
power configuration is preventing link establishment.
Signed-off-by: Jeb Cramer <jeb.j.cramer@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
For FDIR, GTPU with inner L3/L4 layers should only support inner
L3/L4 addrs/ports, instead of outer fields. Thus, we use TUN offsets
for GTPU IP/EH to insert inner L3/L4 addrs/ports fields.
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Add capability which indicates double reset solution restriction.
Added "Post-update EMPR enabled" field to "Response Flags" field
(byte 19 in the response structure).
Signed-off-by: Amir Shay <shay.amir@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
