OCTEON TX2 isn't built for gcc 4.8.5 as the compiler emits
"internal compiler error" on aarch64. This causes the following
build error when OCTEON TX2 EP is enabled:
/usr/bin/ld: cannot find -lrte_common_octeontx2
collect2: error: ld returned 1 exit status
Fixes: 56d46d13f7 ("raw/octeontx2_ep: add build infra and device probe")
Signed-off-by: Ali Alnubani <alialnu@mellanox.com>
In the Rx datapath the flags in the newly allocated mbufs
are all explicitly cleared but the EXT_ATTACHED_MBUF must be
preserved. It would allow to use mbuf pools with pre-attached
external data buffers.
The vectorized rx_burst routines are updated in order to
inherit the EXT_ATTACHED_MBUF from mbuf pool private
RTE_PKTMBUF_POOL_F_PINNED_EXT_BUF flag.
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
The new mbuf pool type is added to testpmd. To engage the
mbuf pool with externally attached data buffers the parameter
"--mp-alloc=xbuf" should be specified in testpmd command line.
The objective of this patch is just to test whether mbuf pool
with externally attached data buffers works OK. The memory for
data buffers is allocated from DPDK memory, so this is not
"true" external memory from some physical device (this is
supposed the most common use case for such kind of mbuf pool).
The user should be aware that not all drivers support the mbuf
with EXT_ATTACHED_BUF flags set in newly allocated mbuf (many
PMDs just overwrite ol_flags field and flag value is getting
lost).
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
The dedicated routine rte_pktmbuf_pool_create_extbuf() is
provided to create mbuf pool with data buffers located in
the pinned external memory. The application provides the
external memory description and routine initializes each
mbuf with appropriate virtual and physical buffer address.
It is entirely application responsibility to register
external memory with rte_extmem_register() API, map this
memory, etc.
The new introduced flag RTE_PKTMBUF_POOL_F_PINNED_EXT_BUF
is set in private pool structure, specifying the new special
pool type. The allocated mbufs from pool of this kind will
have the EXT_ATTACHED_MBUF flag set and initialiazed shared
info structure, allowing cloning with regular mbufs (without
attached external buffers of any kind).
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
Update detach routine to check the mbuf pool type.
Introduce the special internal version of detach routine to handle
the special case of pinned external bufferon mbuf freeing.
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
The routine rte_pktmbuf_priv_flags is introduced to fetch
the flags from the mbuf memory pool private structure
in unified fashion.
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
Use new marker typedef available in EAL and remove private marker
typedef.
Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Use new marker typedef available in EAL.
Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Introduce EAL typedef for structure 1B, 2B, 4B, 8B alignment marking and
a generic marker for a point in a structure.
Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Matan Azrad <matan@mellanox.com>
In case of too low number of memzone segments user notification
was misleading. This patch improves the description by providing
better explanation about the cause.
Signed-off-by: Artur Trybula <arturx.trybula@intel.com>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
FreeBSD 13 has changed the definition of vm_page_replace so we need
to have slightly different code paths around this function depending on
the BSD version.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
In (currently unreleased) FreeBSD 13, the CPU_NAND macro has been renamed
to CPU_ANDNOT, so we need to use different DPDK-specific macros depending
on what system-defined ones are present.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Meson versions 0.52 and 0.53 are being overly smart and detecting the path
"/sys/devices/system/cpu/present" in the call to cat in
app/test/meson.build and then adding it as a dependency to the build
configuration. This causes issues on systems where the timestamp of that
file always returns the current time, since it means that the build.ninja
file is always out of date, and therefore needs to be rebuilt.
We can fix this by just using a simple shell script to return the coremask
appropriately for BSD and Linux, and removing that code logic from meson -
thereby hiding the use of the /sys file.
Fixes: c70622ac6f ("test: detect number of cores with meson")
Cc: stable@dpdk.org
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Moved RFC4115 APIs to non-experimental as they have been there
since 19.02. Also, these APIs are the same as the non RFC4115 APIs.
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Acked-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Enhance API documentation of rte_pktmbuf_attach_extbuf() to
explain that the attached mbuf is initialized with length = 0.
Link: https://bugs.dpdk.org/show_bug.cgi?id=362
Signed-off-by: Jörg Thalheim <joerg@thalheim.io>
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
The existing optimize_object_size() function address the memory object
alignment constraint on x86 for better performance.
Different (micro) architecture may have different memory alignment
constraint for better performance and it not the same as the existing
optimize_object_size().
Some use, XOR(kind of CRC) scheme to enable DRAM channel distribution
based on the address and some may have a different formula.
Introducing arch_mem_object_align() function to abstract
the difference between different (micro) architectures to avoid
wasting memory for mempool object alignment for the architecture
that it is not required to do so.
Details on the amount of memory saving:
Currently, arm64 based architectures use the default (nchan=4,
nrank=1). The worst case is for an object whose size (including mempool
header) is 2 cache lines, where it is optimized to 3 cache lines (+50%).
Examples for cache lines size = 64:
orig optimized
64 -> 64 +0%
128 -> 192 +50%
192 -> 192 +0%
256 -> 320 +25%
320 -> 320 +0%
384 -> 448 +16%
...
2304 -> 2368 +2.7% (~mbuf size)
Additional details:
https://www.mail-archive.com/dev@dpdk.org/msg149157.html
Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
Starting from v20.05, rte_mempool_populate_iova() will return 0.
The ABI will be preserved through symbol versioning until 20.11.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
To populate a mempool with a virtual area, the mempool code calls
rte_mempool_populate_iova() for each iova-contiguous area. It happens
(rarely) that this area is too small to store one object. In this case,
rte_mempool_populate_iova() returns an error, which is forwarded by
rte_mempool_populate_virt().
This case should not throw an error in rte_mempool_populate_virt().
Instead, the area that is too small should just be ignored.
To fix this issue, change the return value of
rte_mempool_populate_iova() to 0 when no object can be populated,
so it can be ignored by the caller. As this would be an API/ABI change,
only do this modification internally for now.
Fixes: 354788b60c ("mempool: allow populating with unaligned virtual area")
Cc: stable@dpdk.org
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Tested-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: Alvin Zhang <alvinx.zhang@intel.com>
When allocating a mempool which is larger than the largest
available area, it can take a lot of time:
a- the mempool calculate the required memory size, and tries
to allocate it, it fails
b- then it tries to allocate the largest available area (this
does not request new huge pages)
c- add this zone to the mempool, this triggers the allocation
of a mem hdr, which request a new huge page
d- back to a- until mempool is populated or until there is no
more memory
This can take a lot of time to finally fail (several minutes): in step
a- it takes all available hugepages on the system, then release them
after it fails.
The problem appeared with commit eba11e3646 ("mempool: reduce wasted
space on populate"), because smaller chunks are now allowed. Previously,
it had to be at least one page size, which is not the case in step b-.
To fix this, implement our own way to allocate the largest available
area instead of using the feature from memzone: if an allocation fails,
try to divide the size by 2 and retry. When the requested size falls
below min_chunk_size, stop and return an error.
Fixes: eba11e3646 ("mempool: reduce wasted space on populate")
Cc: stable@dpdk.org
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
Tested-by: Ali Alnubani <alialnu@mellanox.com>
The documentation says that a negative errno is returned on error, but
in most places that's not the case.
Fix the documentation and the exceptions in code. The second one
(return from populate_virt) also fixes a memory leak.
Note that testpmd was using the function correctly.
Fixes: aa10457eb4 ("mempool: make mempool populate and free api public")
Fixes: 6780f72fb8 ("mempool: populate with anonymous memory")
Fixes: 66e7ba0bad ("mempool: ensure mempool is initialized before populating")
Cc: stable@dpdk.org
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
The rte_security API which enables inline protocol/crypto feature
mandates that for every security session an rte_flow is created. This
would internally translate to a rule in the hardware which would do
packet classification.
In rte_security, one SA would be one security session. And if an rte_flow
need to be created for every session, the number of SAs supported by an
inline implementation would be limited by the number of rte_flows the
PMD would be able to support.
If the fields SPI & IP addresses are allowed to be a range, then this
limitation can be overcome. Multiple flows will be able to use one rule
for SECURITY processing. In this case, the security session provided as
conf would be NULL.
Application should do an rte_flow_validate() to make sure the flow is
supported on the PMD.
Signed-off-by: Anoob Joseph <anoobj@marvell.com>
Reviewed-by: Jerin Jacob <jerinj@marvell.com>
Acked-by: Ori Kam <orika@mellanox.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Acked-by: Vladimir Medvedkin <vladimir.medvedkin@intel.com>
The version of the PMD should not be overwritten with LIBABIVER.
All internal libs must have the same version.
Moreover, specifying LIBABIVER in the driver causes a symbolic link loop:
librte_rawdev_octeontx2_ep.so.1 -> librte_rawdev_octeontx2_ep.so.1
Fixes: 56d46d13f7 ("raw/octeontx2_ep: add build infra and device probe")
Reported-by: Ferruh Yigit <ferruh.yigit@intel.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Tested-by: Ferruh Yigit <ferruh.yigit@intel.com>
Mempool ops may register in different order when compiling. The default
ops index is always zero but the wanted ops is ring_mp_mc. This patch
sets best mempool ops before generating mbuf pool to fix this issue.
Fixes: 5194299d6e ("examples/ntb: support more functions")
Cc: stable@dpdk.org
Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
In prerequisites of ntb guide, the correct flag when loading igb_uio
module should be `wc_activate=1`, not `wc_active=1`.
Fixes: 11b5c7daf0 ("raw/ntb: add enqueue and dequeue functions")
Cc: stable@dpdk.org
Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>
All buffers and ring info should be written before tail register update.
This patch relocates the write memory barrier before updating tail register
to avoid potential issues.
Fixes: 11b5c7daf0 ("raw/ntb: add enqueue and dequeue functions")
Cc: stable@dpdk.org
Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>
Add rawdev's selftest feature in SDP VF driver, which
verifies the EP mode functionality test.
Signed-off-by: Mahipal Challa <mchalla@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Register "dev_configure" API to configure/initialize the SDP
VF PCIe devices.
Signed-off-by: Mahipal Challa <mchalla@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Add the OCTEON TX2 SDP EP device probe along with the
build infrastructure for Make and meson builds.
Signed-off-by: Mahipal Challa <mchalla@marvell.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
It is useful to know when the next timer will expire when
using rte_epoll_wait (or sleep when idle). This experimental
API provides a hook to query the number of ticks remaining.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Erik Gabriel Carrillo <erik.g.carrillo@intel.com>
Make latency calculation multithread safe by
using spinlock.
Fixes: 5cd3cac9ed ("latency: added new library for latency stats")
Cc: stable@dpdk.org
Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
Adding new API function to query the maximum key ID
that could possibly be returned by rte_hash_add_key and
rte_hash_add_key_with_hash. When RTE_HASH_EXTRA_FLAGS_MULTI_WRITER_ADD
is set, the maximum key id is larger than the entry count specified
by the user.
Signed-off-by: Kumar Amber <kumar.amber@intel.com>
Acked-by: Yipeng Wang <yipeng1.wang@intel.com>
rte_cfgfile_section_num_entries_by_index was missing from the map file.
meson build failed when calling this function,
due to linking a binary to cfgfile built as a shared library.
Fixes: 3d2e0448eb ("cfgfile: add section number of entries by index")
Cc: stable@dpdk.org
Signed-off-by: Liron Himi <lironh@marvell.com>
Acked-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
All global variables in kernel should be prefixed by the same
to avoid any symbol conflics. Rename dflt_carrier to kni_default_carrier.
Fixes: 89397a01ce ("kni: set default carrier state of interface")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
The header linux/version.h isn't included when CONFIG_RTE_EAL_VFIO
is explicitly disabled. LINUX_VERSION_CODE and KERNEL_VERSION are
therefore undefined, causing the build failure:
lib/librte_eal/linux/eal/eal.c: In function ‘rte_eal_init’:
lib/librte_eal/linux/eal/eal.c:1076:32: error: "LINUX_VERSION_CODE" is
not defined, evaluates to 0 [-Werror=undef]
Fixes: a0dede62a5 ("eal/linux: remove KNI restriction on IOVA")
Cc: stable@dpdk.org
Signed-off-by: Ali Alnubani <alialnu@mellanox.com>
Running ld with -r switches the linker to a very special mode where
some other linker options don't make sense.
In particular, -export-dynamic normally requires that all global
symbols be included in the dynamic symbol table, but a .o file doesn't
even have a dynamic symbol table.
When given both options it looks like the gnu linker just ignores
-export-dynamic.
Unfortunately some versions of lld (https://lld.llvm.org/) have a bug
that causes it to try to create a dynamic symbol table in the output
.o file and ends up corrupting it
(https://bugs.llvm.org/show_bug.cgi?id=43552). Current (git) version
of lld now issues an error.
This patch filters out -export-dynamic from $(LDFLAGS) when using
-r. With this patch I can build dpdk with lld.
Fixes: 3d781ca328 ("mk: do post processing on objects that register a driver")
Cc: stable@dpdk.org
Signed-off-by: Rafael Ávila de Espíndola <espindola@scylladb.com>
Add pmd unix socket server to enable external tool applications to
trigger flow dump.
Socket path:
/var/tmp/dpdk_mlx5_<pid>
Socket format:
io_raw: port_id of uint16
file: file descriptor of int
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Signed-off-by: Xiaoyu Min <jackmin@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
New flow dump CLI to dump device internal representation information
of flows into screen.
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Signed-off-by: Xiaoyu Min <jackmin@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
Dump fdb/nic_rx/nic_tx raw flow data into specified file.
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Signed-off-by: Xiaoyu Min <jackmin@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Introduce an API which dump the device's internal representation
information of rte flows in hardware.
Signed-off-by: Xiaoyu Min <jackmin@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Avoid overwriting device flags and other information in device
data stored in shared memory when a secondary process
probes PCI device.
Fixes: 494adb7f63 ("ethdev: add device fields from PCI layer")
Cc: stable@dpdk.org
Signed-off-by: Fang TongHao <fangtonghao@sangfor.com.cn>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
The mlx5 datapath does not implement any endianness conversions
for the metadata being sent and received to provide the better
performance (because these conversions would be performed for
each packet). These metadata are also involved into flow processing
(there might be some flows matching on metadata patterns or setting
the new metadata values) inside the NIC. It order to configure
hardware in correct way all necessary endianness conversions are
done by rte_flow handling code (only once on flow creation). This
patch fixes one of these conversions for the little-endian hosts
in case if META/MARK items are less than 32 bits.
Fixes: acfcd5c52f ("net/mlx5: update meta register matcher set")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Matcher is flow table related structure providing the flow pattern
to be translated directly in hardware controlling data. This structure
includes the metadata register c0 field, that might be engaged to
support META and MARK related flow items and actions. Also, this
register might be used by kernel to specify the source vport index.
In this case (if kernel uses the field) the register c0 is split
into two 16-bit subfields - one for META/MARK items and another
to handle vport.
The actual configuration is queried by PMD from kernel in runtime
and depending on the mask returned by kernel the PMD can use upper
or lower half of register c0 field. This patch adds the missing
support for upper half. This missed support caused the non-operational
META/MARK items on some kernel configurations.
Fixes: e554b672aa ("net/mlx5: support flow tag")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Matcher is flow table related structure providing the flow pattern
to be translated directly in hardware controlling data. Some fields
in this structure might be split (by software) between multiple items.
For example, the metadata register c0 field in the matcher might be
split into two independent subfields - the source vport index and
META item value. These subfields have no permanent assigned masks,
the actual configuration is queried from the kernel drivers in
runtime. To handle source vport value (the port of e-Switch which
is origin of the packet) the kernel might use the dedicated vport
field in the matcher or the part of register c0 field, depending
on configuration.
To setup the matcher structure fields the macro MLX5_SET is used.
MLX5_SET configures the specified 32-bit field as whole entity.
For metadata register c0 we should take into account the provided
mask in order to configure the specified subfield bits only,
otherwise setting vport overrides the META values and vice versa.
Fixes: acfcd5c52f ("net/mlx5: update meta register matcher set")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Currently, there is a potential problem that changing the content of
dev->data->dev_conf.rxmode.offloads even when there is no
vlan_offload_set driver callback.
It is a good idea that prevent the side effect and make the API return
success if no change requested. This patch fixes the problem, the detail
information as below:
- keep possibility to do dummy set even if there is no driver callback
- do not touch Rx mode offloads in device data before checking the
driver callback availability
- ensure that Rx mode offloads are rolled back correctly if driver
callback returns error
Fixes: 81f9db8ecc ("ethdev: add vlan offload support")
Cc: stable@dpdk.org
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Signed-off-by: Chunsong Feng <fengchunsong@huawei.com>
Signed-off-by: Min Wang (Jushui) <wangmin3@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
