net/mlx5: fix actions validation on root table

The maximal supported header modifications number of a single modify
context on the root table cannot be queried from firmware directly.
It is a fixed value of 16 in the latest releases. In the validation
stage, PMD driver should ensure that no more than 16 header modify
actions exist in a single context.
In some old firmware releases, the supported value is 8. PMD driver
should try its best to create the flow. Firmware will return error
and refuse to create the flow if the actions number exceeds the
maximal value.

Fixes: 72a944dba163 ("net/mlx5: fix header modify action validation")
Cc: stable@dpdk.org
Signed-off-by: Bing Zhao <bingz@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

net/mlx5: fix indexed pool bitmap initialization

Currently, the indexed memory pool bitmap start address is not aligned
to cacheline size explicitly. The bitmap initialization requires the
address should be cacheline aligned. In that case, the initialization
maybe failed if the address is not cacheline aligned.

Add RTE_CACHE_LINE_ROUNDUP() to the trunk size calculation to make sure
the bitmap offset address will start with cacheline aligned.

Fixes: a3cf59f56c47 ("net/mlx5: add indexed memory pool")

Signed-off-by: Suanming Mou <suanmingm@mellanox.com>
Tested-by: Lijian Zhang <lijian.zhang@arm.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>

net/mlx5: fix packet length assert in MPRQ

The assert that checks if there is a enough room for the
whole packet minus headroom data is written incorrectly.
The check should be negated in order to work properly.

Fixes: bd0d5930bf56 ("net/mlx5: enable MPRQ multi-stride operations")

Signed-off-by: Alexander Kozyrev <akozyrev@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

net/mlx5: fix assert in dynamic metadata handling

The assert in dynamic flow metadata handling is wrong after the
fix for the performance degradation. The assert meant to check
the metadata mask but was updated with the metadata offset instead.
Fix this assert and restore proper metadata mask checking.

Fixes: 6c55b622a956 ("net/mlx5: set dynamic flow metadata in Rx queues")
Cc: stable@dpdk.org
Signed-off-by: Alexander Kozyrev <akozyrev@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

vhost: use binary search in address conversion

If Tx zero copy enabled, gpa to hpa mapping table is updated one by
one. This will harm performance when guest memory backend using 2M
hugepages. Now utilize binary search to find the entry in mapping
table, meanwhile set the threshold to 256 entries for linear search.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: utilize dynamic memory allocator

Replace dynamic memory allocator with dpdk memory allocator.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: prevent zero-copy with incompatible client mode

In server mode, virtio-user inits under the assumption that vhost-user
supports a list of features. However, this could be problematic when
in_order feature is negotiated but not supported by vhost-user when
enables dequeue_zero_copy later.

Add handling when vhost-user enables dequeue_zero_copy as client.

Fixes: 64ab701c3d1e ("vhost: add vhost-user client mode")
Cc: stable@dpdk.org
Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Tested-by: Yinan Wang <yinan.wang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: add election for vectorized path

Rewrite vectorized path selection logic. Default setting comes from
vectorized devarg, then checks each criteria.

Packed ring vectorized path need:
    AVX512F and required extensions are supported by compiler and host
    VERSION_1 and IN_ORDER features are negotiated
    mergeable feature is not negotiated
    LRO offloading is disabled

Split ring vectorized rx path need:
    mergeable and IN_ORDER features are not negotiated
    LRO, chksum and vlan strip offloadings are disabled

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: add vectorized packed ring Tx

Optimize packed ring Tx path like Rx path. Split Tx path into batch and
single Tx functions. Batch function is further optimized by AVX512
instructions.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: add vectorized packed ring Rx

Optimize packed ring Rx path with SIMD instructions. Solution of
optimization is pretty like vhost, is that split path into batch and
single functions. Batch function is further optimized by AVX512
instructions.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: reuse packed ring functions

Move offload, xmit cleanup and packed xmit enqueue function to header
file. These functions will be reused by packed ring vectorized path.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio-user: add parameter to enable vectorized path

Add new devarg for virtio user device vectorized path selection.
By default vectorized path is disabled.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: add parameter to enable vectorized path

Previously, virtio split ring vectorized path was enabled by default.
This is not suitable for everyone because that path does not follow
virtio spec. Add new devarg for virtio vectorized path selection. By
default vectorized path is disabled.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: enable in-order feature if negotiated

Ring initialization is different when inorder feature negotiated. This
action should dependent on negotiated feature bits.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: add Rx free threshold setting

Introduce free threshold setting in Rx queue, its default value is 32.
Limit the threshold size to multiple of four as only vectorized packed
Rx function will utilize it. Virtio driver will rearm Rx queue when
more than rx_free_thresh descs were dequeued.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vdpa/mlx5: recreate a virtq becoming enabled

The virtq configurations may be changed when it moves from disabled
state to enabled state.

Listen to the state callback even if the device is not configured.
Recreate the virtq when it moves from disabled state to enabled state
and when the device is configured.

Signed-off-by: Matan Azrad <matan@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vdpa/mlx5: separate virtq stop

In live migration, before logging the virtq, the driver queries the
virtq indexes after moving it to suspend mode.

Separate this method to new function mlx5_vdpa_virtq_stop as a
preparation for reusing.

Signed-off-by: Matan Azrad <matan@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vdpa/mlx5: manage virtqs by array

As a preparation to listen the virtqs status before the device is
configured, manage the virtqs structures in array instead of list.

Signed-off-by: Matan Azrad <matan@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: optimize broadcast RARP sync with C11 atomic

The rarp packet broadcast flag is synchronized with rte_atomic_XX APIs
which is a full barrier, DMB, on aarch64. This patch optimized it with
c11 atomic one-way barrier.

Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Joyce Kong <joyce.kong@arm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: fix peer close check

In process_slave_message_reply(), there is a
possibility that receiving a peer close
message instead of a real message response.

This patch targeting to handle the peer close
scenario and report the correct error message.

Fixes: a277c7159876 ("vhost: refactor code structure")
Cc: stable@dpdk.org
Signed-off-by: Roland Qi <roland.qi@ucloud.cn>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

doc: remove virtio vector PMD from NIC features table

The goal is to make the table more readable.

Mark as partially supported features that are supported in
the generic Virtio driver but not in the vectorized ones.

Suggested-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/mlx5: save meter index instead of meter id

Currently, while creating the flow with meter, meter id is saved to the
rte flow. While destroying the flow, the meter object will be found by
the meter id, so the meter object will be released accordingly. But as
the meter id is configured by user, while the meter id is set to 0, it
doesn't make any sense to flow destroy since 0 means flow doesn't have
meter. The meter object with id 0 will be leaked.

As meter object is allocated from indexed memory, and the index starts
from 1, save the internal generated index instead of user defined meter
id will never meet the issue as above.

This patch saves meter index instead of meter id in rte flow.

Signed-off-by: Suanming Mou <suanmingm@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

common/mlx5: fix uninitialized RoCE variable

Gcc 8.3.0 (Debian 10) complains about uninitialized variable.

[474/2122] Compiling C object
'drivers/a715181@@tmp_rte_common_mlx5@sta/common_mlx5_mlx5_nl.c.o'.
In file included from ../drivers/common/mlx5/mlx5_nl.h:12,
                 from ../drivers/common/mlx5/mlx5_nl.c:23:
../drivers/common/mlx5/mlx5_nl.c: In function ‘mlx5_nl_enable_roce_get’:
../drivers/common/mlx5/mlx5_common.h:68:2: warning: ‘cur_en’ may be used
uninitialized in this function [-Wmaybe-uninitialized]
  rte_log(RTE_LOG_ ## level, \
  ^~~~~~~
../drivers/common/mlx5/mlx5_nl.c:1560:6: note: ‘cur_en’ was declared here
  int cur_en;
      ^~~~~~

The compiler is correct, this variable would only be set if kernel
netlink response message contains the DEVLINK parameter that flags if
ROCE is enabled.

Fixes: fa69eaef5f49 ("common/mlx5: support ROCE disable through Netlink")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

net/mlx5: fix assert in doorbell lookup

The asserts makes sure that 'i' doesn't exceed the expected value.
This to prevent an out of bound access to dbr_bitmap.

The current location of the assert protects the assignment of
dbr_bitmap, but not the access to it.

Moved the assert to the correct place, to protect both cases.
Also, used an existing define for the assert.

Fixes: 21cae8580fd0 ("net/mlx5: allocate door-bells via DevX")
Cc: stable@dpdk.org
Signed-off-by: Asaf Penso <asafp@mellanox.com>
Reviewed-by: Dekel Peled <dekelp@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

net/mlx5: fix empty flow error structure

The output flow error parameter is used to indicate the detailed
reason of the failure when calling a rte_flow_* interface. Even
though sometimes the application will not check it or use it, the PMD
must fill it in the failure branch before returning. Or else, some
dirty value in the stack, heap will be accessed as a pointer and then
cause a crash.
In this case, when a port is stopped, it is not allowed to insert a
flow from application. The detailed error information should be
filled. If the application needs to check the detailed error reason,
it will get the information but not result in any crash.

Fixes: 40b9e7f65fe1 ("net/mlx5: check device status before creating flow")

Signed-off-by: Bing Zhao <bingz@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>

net/mlx5: fix Rx queue flags on destroying flow

After inserting an offload flow, the software flag information will
be updated based on the flow. When receiving a packet on this queue,
the hardware packet type bits and the software flag will be used
together to get the inner packet and tunnel header type (if any) from
the global packet type table.
When destroying a flow, the corresponding Rx queue flag needs to be
updated. All flags should be cleared when closing a device because
all control flows and application flows are invalid anymore.
Such behavior is missed when implementing the non-cached mode.

Fixes: 8db7e3b69822 ("net/mlx5: change operations for non-cached flows")

Signed-off-by: Bing Zhao <bingz@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>

net/ice/base: fix DCF switch rule

1. ln_en bit should not be turned on, since we only support Rx VEB.
2. lan_en bit need to be turned on for a DCF switch rule, otherwise
   any Tx packet that hit on a rule will be dropped.

Fixes: fed0c5ca5f19 ("net/ice/base: support programming a new switch recipe")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>

net/iavf: fix VF reset for flow director rule

After VF reset, FDIR rule still takes effect. To solve the issue,
this patch adds to flush all flows before flow uninit. VIRTCHNL
sends message to PF by Admin Queue, so flow flush should be implemented
before Admin Queue shut down.

Fixes: ff2d0c345c3b ("net/iavf: support generic flow API")

Signed-off-by: Simei Su <simei.su@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>

net/i40e: fix flush of flow director filter

When we flush FDIR filter, we can not call i40e_fdir_teardown()
function as it will free vsi used for FDIR, then the vsi->base_queue
will be freed from pf->qp_pool, but vsi->base_queue can only get
once when do dev init in i40e_pf_setup(). If we free it, it will
never be alloc again.

Bugzilla ID: 404
Fixes: 2e67a7fbf3ff ("net/i40e: config flow director automatically")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

doc: add i40e limitation for flow director

Each PCTYPE can only have one specific FDIR input set at one time.
Add input set requirement info to i40e doc.

Bugzilla ID: 403
Fixes: 14c66a451ef7 ("net/i40e: flush tunnel filters")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e: fix flow director for ARP packets

Currently, flow "pattern eth type is 0x0806 / end actions mark id
0x86 / rss / end" can't be created successfully. FDIR parser
shouldn't deny RTE_ETHER_TYPE_ARP since ARP packets will be
parsed as PCTYPE_L2_PAYLOAD. This patch fixes the issue.

Bugzilla ID: 402
Fixes: 42044b69c67d ("net/i40e: support input set selection for FDIR")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/ice/base: workaround unexpected rule deletion

Ideally a rule with "TO VSI LIST" action should not be deleted when one
of the VF reset happens. The correct action by kernel PF driver is to
remove the VSI of a reset VF from the VSI list, but this is not
implemented in kernel PF yet, so workaround is the DCF to prevent a
rule with "To VSI List" action happens.

Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Reviewed-by: Wei Zhao <wei.zhao1@intel.com>

net/bnxt: add Truflow flush-timer to alloc table scope

Updated the params list to include flush timer, this will
allow users to set the HW flush timer value in 10th of second.
Setting 0 will disable the pending cache flush feature.

Signed-off-by: Shahaji Bhosle <sbhosle@broadcom.com>
Signed-off-by: Randy Schacher <stuart.schacher@broadcom.com>
Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>

net/bnxt: update ULP to handle action/index tables

The ulp required to be changed to properly call the index table
management routines and use the index for external memory indices.
The ulp no longer has to account for stride as the tf_core returns the
actual offset, not a 0 based index.

Signed-off-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Kishore Padmanabha <kishore.padmanabha@broadcom.com>
Reviewed-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>

net/bnxt: update external action record pool

- Added support variable sized action records
- Additional error checking on table scope params
- Single external pool supported per direction
- Changed to return action record pointer
- Allows action pool to fully utilize the number of flows

Signed-off-by: Farah Smith <farah.smith@broadcom.com>
Signed-off-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Peter Spreadborough <peter.spreadborough@broadcom.com>
Reviewed-by: Kishore Padmanabha <kishore.padmanabha@broadcom.com>

net/bnxt: enable processing partially added flows

The resource function did not have a method of invalidating or
indicating that a resource is uninitialized.  Added an invalid enum so
that processing works correctly for partially added flows.

Signed-off-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Kishore Padmanabha <kishore.padmanabha@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>

net/bnxt: fix max ring count

Max Rx Ring count could be < Max stat contexts. While accounting
for stat contexts, this should be also considered and
the max ring count adjusted accordingly.

Fixes: f03e66cb64ce ("net/bnxt: limit queue count for NS3/Stingray devices")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>

eal/ppc: fix build with gcc 9.3

Building DPDK on Ubuntu 20.04 with GCC 9.3.0 results in a "subscript is
outside array bounds" message in rte_memcpy function.  The build error
is caused by an interaction between __builtin_constant_p and
"-Werror=array-bounds" as described in this bugzilla:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90387

Modify the code to disable the array-bounds check for GCC versions 9.0
to 9.3.

Cc: stable@dpdk.org
Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>

kvargs: fix invalid token parsing on FreeBSD

The behavior of strtok_r() is not the same between GNU libc and FreeBSD
libc: in the first case, the context is set to "" when the last token is
returned, while in the second case it is set to NULL.

On FreeBSD, the current code crashes because we are dereferencing a NULL
pointer (ctx1). Fix it by first checking if it is NULL. This works with
both GNU and FreeBSD libc.

Fixes: ffcf831454a9 ("kvargs: fix buffer overflow when parsing list")
Cc: stable@dpdk.org
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Tested-by: Zhimin Huang <zhiminx.huang@intel.com>

trace: fix build with gcc 10

Prevent from writing beyond the allocated memory.

GCC 10 compiling output:
eal_common_trace_utils.c: In function 'eal_trace_dir_args_save':
eal_common_trace_utils.c:290:24: error: '__builtin___sprintf_chk'   \
may write a terminating nul past the end of the destination \
[-Werror=format-overflow=]
  290 |  sprintf(dir_path, "%s/", optarg);
      |                        ^

Fixes: 8af866df8d8c ("trace: add trace directory configuration parameter")

Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Lijian Zhang <lijian.zhang@arm.com>
Tested-by: Lijian Zhang <lijian.zhang@arm.com>
Acked-by: Sunil Kumar Kori <skori@marvell.com>
Signed-off-by: David Marchand <david.marchand@redhat.com>

trace: remove string duplication

No need to duplicate an untouched string.

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Sunil Kumar Kori <skori@marvell.com>

trace: remove limitation on patterns number

There is nothing performance sensitive in this list, use dynamic
allocations and remove the arbitrary limit on the number of trace
patterns a user can pass.

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Sunil Kumar Kori <skori@marvell.com>

trace: remove unneeded checks in internal API

The trace framework can be configured via 4 EAL options:
- --trace which calls eal_trace_args_save,
- --trace-dir which calls eal_trace_dir_args_save,
- --trace-bufsz which calls eal_trace_bufsz_args_save,
- --trace-mode which calls eal_trace_mode_args_save.

Those 4 internal callbacks are getting passed a non NULL value:
optarg won't be NULL since those options are declared with
required_argument (man getopt_long).

eal_trace_bufsz_args_save() already trusted passed value, align the other
3 internal callbacks.

Coverity issue: 357768
Fixes: 8c8066ea6a7b ("trace: add trace mode configuration parameter")

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Sunil Kumar Kori <skori@marvell.com>

trace: avoid confusion on optarg

Prefer a local name to optarg which is a global symbol from the C library.

Fixes: 8c8066ea6a7b ("trace: add trace mode configuration parameter")

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Sunil Kumar Kori <skori@marvell.com>

trace: simplify trace point headers

Invert the current trace point headers logic by making
rte_trace_point_register.h include rte_trace_point.h.

There is no more need for a RTE_TRACE_POINT_REGISTER_SELECT special macro
since including rte_trace_point_register.h itself means we want to
register trace points.

The unexplained "provider" notion is removed from the documentation and
rte_trace_point_provider.h is merged into rte_trace_point.h.

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>

cryptodev: fix trace points registration

Those trace points are defined but not registered.

Fixes: 4cf30e3f3c35 ("cryptodev: add tracepoints")

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Sunil Kumar Kori <skori@marvell.com>

eal/ppc: fix bool type after altivec include

The AltiVec header file breaks boolean type. [1] [2]

Currently the workaround was located only in mlx5 device.
Adding the trace module caused this issue to appear again, due to
order of includes, it keeps overriding the local fix.

This patch solves this issue by resetting the bool type, immediately
after it is being changed.

[1] https://mails.dpdk.org/archives/dev/2018-August/110281.html

[2]
In file included from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:18:0,
                 from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool.h:54,
                 from
dpdk/drivers/common/mlx5/mlx5_common_mr.c:7:
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h: In
function '__rte_trace_point_fp_is_enabled':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:226:2:
error: incompatible types when returning type 'int' but '__vector __bool
int' was expected
  return false;
  ^
In file included from
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:281:0,
                 from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:18,
                 from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool.h:54,
                 from
dpdk/drivers/common/mlx5/mlx5_common_mr.c:7:
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:
In function 'rte_mempool_trace_ops_dequeue_bulk':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point_provider.h:104:6:
error: wrong type argument to unary exclamation mark
  if (!__rte_trace_point_fp_is_enabled()) \
      ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:49:2:
note: in expansion of macro '__rte_trace_point_emit_header_fp'
  __rte_trace_point_emit_header_##_mode(&__##_tp); \
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:99:2:
note: in expansion of macro '__RTE_TRACE_POINT'
  __RTE_TRACE_POINT(fp, tp, args, __VA_ARGS__)
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:20:1:
note: in expansion of macro 'RTE_TRACE_POINT_FP'
 RTE_TRACE_POINT_FP(
 ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:
In function 'rte_mempool_trace_ops_dequeue_contig_blocks':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point_provider.h:104:6:
error: wrong type argument to unary exclamation mark
  if (!__rte_trace_point_fp_is_enabled()) \
      ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:49:2:
note: in expansion of macro '__rte_trace_point_emit_header_fp'
  __rte_trace_point_emit_header_##_mode(&__##_tp); \
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:99:2:
note: in expansion of macro '__RTE_TRACE_POINT'
  __RTE_TRACE_POINT(fp, tp, args, __VA_ARGS__)
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:29:1:
note: in expansion of macro 'RTE_TRACE_POINT_FP'
 RTE_TRACE_POINT_FP(
 ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:
In function 'rte_mempool_trace_ops_enqueue_bulk':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point_provider.h:104:6:
error: wrong type argument to unary exclamation mark
  if (!__rte_trace_point_fp_is_enabled()) \

Fixes: 725f5dd0bfb5 ("net/mlx5: fix build on PPC64")

Signed-off-by: Ori Kam <orika@mellanox.com>
Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Tested-by: David Christensen <drc@linux.vnet.ibm.com>
Tested-by: Raslan Darawsheh <rasland@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>

eal/x86: ignore gcc 10 stringop-overflow warnings

stringop-overflow warns when it sees a possible overflow
in a string operation.

In the rte_memcpy functions different branches are taken
depending on the size. stringop-overflow is raised for the
branches in the function where it sees the static size of the
src could be overflowed.

However, in reality a correct size argument and in some cases
dynamic allocation would ensure that this does not happen.

For example, in the case below for key, the correct path will be
chosen in rte_memcpy_generic at runtime based on the size argument
but as some paths in the function could lead to a cast to 32 bytes
a warning is raised.

In function ‘_mm256_storeu_si256’,
inlined from ‘rte_memcpy_generic’
at ../lib/librte_eal/common/include/arch/x86/rte_memcpy.h:315:2,
inlined from ‘iavf_configure_rss_key’
at ../lib/librte_eal/common/include/arch/x86/rte_memcpy.h:869:10:

/usr/lib/gcc/x86_64-redhat-linux/10/include/avxintrin.h:928:8:
warning: writing 32 bytes into a region of size 1 [-Wstringop-overflow=]
  928 |   *__P = __A;
      |   ~~~~~^~~~~
In file included
from ../drivers/net/iavf/../../common/iavf/iavf_prototype.h:10,
from ../drivers/net/iavf/iavf.h:9,
from ../drivers/net/iavf/iavf_vchnl.c:22:

../drivers/net/iavf/iavf_vchnl.c:
In function ‘iavf_configure_rss_key’:

../drivers/net/iavf/../../common/iavf/virtchnl.h:508:5:
note: at offset 0 to object ‘key’ with size 1 declared here
  508 |  u8 key[1];         /* RSS hash key, packed bytes */
      |     ^~~

Ignore the stringop-overflow warnings for rte_memcpy.h functions.

Bugzilla ID: 394
Bugzilla ID: 421
Cc: stable@dpdk.org
Signed-off-by: Kevin Traynor <ktraynor@redhat.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

examples/ipsec-gw: fix gcc 10 maybe-uninitialized warning

gcc 10.0.1 reports:

../examples/ipsec-secgw/ipsec_process.c: In function ‘ipsec_process’:
../examples/ipsec-secgw/ipsec_process.c:132:34:
error: ‘grp.m’ may be used uninitialized in this function
 [-Werror=maybe-uninitialized]
  132 |    grp[n].cnt = pkts + i - grp[n].m;
      |                            ~~~~~~^~

This is a correct warning for the initial execution of the statement.
However, it is the design of the loop that grp[0].cnt will later be
written with the correct value using an initialized grp[0].m before it
is used.

In order to remove the warning, initialize grp[0].m for the initial and
unused calculation of grp[0].cnt.

Fixes: 3e5f4625dc17 ("examples/ipsec-secgw: make data-path to use IPsec library")
Cc: stable@dpdk.org
Suggested-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Signed-off-by: Kevin Traynor <ktraynor@redhat.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

net/avp: fix gcc 10 maybe-uninitialized warning

gcc 10.0.1 reports:

../drivers/net/avp/avp_ethdev.c: In function ‘avp_xmit_scattered_pkts’:
../drivers/net/avp/avp_ethdev.c:1791:24:
warning: ‘avp_bufs[count]’ may be used uninitialized in this function
 [-Wmaybe-uninitialized]
 1791 |   tx_bufs[i] = avp_bufs[count];
      |                ~~~~~~~~^~~~~~~
../drivers/net/avp/avp_ethdev.c:1791:24:
warning: ‘avp_bufs[count]’ may be used uninitialized in this function
 [-Wmaybe-uninitialized]

Fix by initializing the array.

Fixes: 295abce2d25b ("net/avp: add packet transmit functions")
Cc: stable@dpdk.org
Signed-off-by: Kevin Traynor <ktraynor@redhat.com>
Acked-by: Steven Webster <steven.webster@windriver.com>

doc: add l3fwd-graph application user guide

Adding the user guide for l3fwd graph application.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

doc: add graph library guide

Adding programmer's guide for Graph library and the inbuilt nodes.
This patch also updates the release note for the new libraries.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>

examples/l3fwd-graph: add graph config and main loop

Add graph creation, configuration logic and graph main loop.
This graph main loop is run on every slave lcore and calls
rte_graph_walk() to walk over lcore specific rte_graph.
Master core accumulates and prints graph walk stats of all the
lcore's graph's.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

examples/l3fwd-graph: add ethdev configuration changes

Add changes to ethdev port and queue configuration based
on command line parameters for l3fwd graph application.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

examples/l3fwd-graph: add graph-based l3fwd skeleton

Add graph based l3fwd application skeleton with cmdline
parsing support inline with normal l3fwd.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

node: add packet drop

Add packet drop node process function for pkt_drop
rte_node. This node simply free's every object received as
an rte_mbuf to its rte_pktmbuf pool.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

node: add IPv4 rewrite and lookup control

Add ip4_rewrite and ip4_lookup ctrl API. ip4_lookup ctrl
API is used to add route entries for LPM lookup with
result data containing next hop id and next proto.
ip4_rewrite ctrl API is used to add rewrite data for
every next hop.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

node: add IPv4 rewrite

Add ip4 rewrite process function for ip4_rewrite
rte_node. On every packet received by this node,
header is overwritten with new data before forwarding
it to next node. Header data to overwrite with is
identified by next hop id passed in mbuf priv data
by previous node.

Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

node: add IPv4 lookup for x86

Add IPv4 lookup process function for ip4_lookup
rte_node. This node performs LPM lookup using x86_64
vector supported RTE_LPM API on every packet received
and forwards it to a next node that is identified by
lookup result.

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

node: add IPv4 lookup for arm64

Add arm64 specific IPv4 lookup process function
for ip4_lookup node. This node performs LPM lookup
on every packet received and forwards it to a next
node that is identified by lookup result.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>

node: add generic IPv4 lookup

Add IPv4 lookup process function for ip4_lookup node.
This node performs LPM lookup using simple RTE_LPM API on every packet
received and forwards it to a next node that is identified by lookup
result.

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

node: add ethdev control

Add ctrl api to setup ethdev_rx and ethdev_tx node.
This ctrl api clones 'N' number of ethdev_rx and ethdev_tx
nodes with specific (port, queue) pairs updated in their context.
All the ethdev ports and queues are setup before this api
is called.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

node: add ethdev Tx

Add rte_node ethdev_tx process function and register it to
graph infra. This node has a specific (port, tx-queue) as context
and it enqueue's all the packets received to that specific queue pair.
When rte_eth_tx_burst() i.e enqueue to queue pair fails, packets
are forwarded to pkt_drop node to be free'd.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

node: add ethdev Rx

Add source rte_node ethdev_rx process function and register
it. This node is a source node that will be called periodically
and when called, performs rte_eth_rx_burst() on a specific
(port, queue) pair and enqueue them as stream of objects to
next node.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

node: add logging and null node

Add log infra for node specific logging.
Also, add null rte_node that just ignores all the objects
directed to it.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>

test/graph: add performance tests

Add unit test framework to create and test performance of various
graph models.

example command to test:

echo "graph_perf_autotest" | sudo ./build/app/test/dpdk-test -c 0x30

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

test/graph: add functional tests

Adding the unit test to test the functionality of node and graph APIs.
Testing includes registering a node, cloning a node, creating a graph,
perform graph walk, collecting stats and all node and graph debug APIs.

example command to test:
echo "graph_autotest" | sudo ./build/app/test/dpdk-test -c 0x30

Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement fastpath routines

Adding implementation for rte_graph_walk() API. This will perform a walk
on the circular buffer and call the process function of each node
and collect the stats if stats collection is enabled.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement stats

Adding implementation for graph stats collection API. This API will
create a cluster for a specified node pattern and aggregate the node
runtime stats.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement debug routines

Adding implementation for graph specific API to dump the
Graph information to a file. This API will dump detailed internal
info about node objects and graph objects.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement Graphviz export

Adding API implementation support exporting the graph object to file.
This will export the graph to a file in Graphviz format.
It can be viewed in many viewers such as
https://dreampuf.github.io/GraphvizOnline/

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement graph operations

Adding support for graph specific API implementation like
Graph lookup to get graph object, retrieving graph ID
From name and graph name from ID.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement create and destroy

Adding graph specific API implementations like graph create
and graph destroy. This detect loops in the graph,
check for isolated nodes and operation to verify the validity of
graph.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: populate fastpath memory for graph reel

Adding support to create and populate the memory for graph reel.
This includes reserving the memory in the memzone, populating the nodes,
Allocating memory for node-specific streams to hold objects.

Once it is populated the reel memory contains the following sections.

+---------------------+
|   Graph Header      |
+---------------------+
|   Fence             |
+---------------------+
|   Circular buffer   |
+---------------------+
|   Fence             |
+---------------------+
|   Node Object 0     |
+------------------- -+
|   Node Object 1     |
+------------------- -+
|   Node Object 2     |
+------------------- -+
|   Node Object n     |
+------------------- -+

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement internal operation helpers

Adding internal graph API helpers support to check whether a graph has
isolated nodes and any node have a loop to itself and BFS
algorithm implementation etc.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement node debug routines

Adding node debug API implementation support to dump
single or all the node objects to the given file.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement node operations

Adding node-specific API implementation like cloning node, updating
edges for the node, shrinking edges of a node, retrieving edges of a
node.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: implement node registration

Adding rte_node_register() API implementation includes allocating
memory for node object, check for duplicate node name and
add the allocated node to STAILQ node_list for future use.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

graph: define API

Graph architecture abstracts the data processing functions as
"node" and "link" them together to create a complex "graph" to enable
reusable/modular data processing functions.

These APIs enables graph framework operations such as create, lookup,
dump and destroy on graph and node operations such as clone,
edge update, and edge shrink, etc. The API also allows creating the
stats cluster to monitor per graph and per node stats.

This patch defines the public API for graph support.
This patch also adds support for the build infrastructure and
update the MAINTAINERS file for the graph subsystem.

Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>

mempool: return 0 if area is too small on populate

Change rte_mempool_populate_iova() and rte_mempool_populate_virt() to
return 0 instead of -EINVAL when there is not enough room to store one
object, as it can be helpful for applications to distinguish this
specific case.

As this is an ABI change, use symbol versioning to preserve old
behavior for binary applications.

Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Lukasz Wojciechowski <l.wojciechow@partner.samsung.com>

version: reference next ABI 21 for recent additions

Change references to ABI 20.0.1 to use ABI v21, see
https://doc.dpdk.org/guides/contributing/abi_policy.html#general-guidelines

"Major ABI versions are declared no more frequently than yearly.
Compatibility with the major ABI version is mandatory in subsequent
releases until a new major ABI version is declared."

Combined ABI policy and versioning in maintainers, add map files to the
filter to more closely monitor future ABI changes.

Signed-off-by: Ray Kinsella <mdr@ashroe.eu>

maintainers: update for testpmd

Replace Jingjing Wu with Beilei Xing.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

maintainers: update for Intel iavf

Replace Wenzhuo Lu with Beilei Xing.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

maintainers: update for Intel i40e

Replace Qi Zhang with Jeff Guo.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Jeff Guo <jia.guo@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

maintainers: update for Intel ixgbe/igb/igc

Replace Wenzhuo Lu, Alvin Zhang and Konstantin Ananyev
with Wei Zhao and Jeff Guo.

Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
Acked-by: Jeff Guo <jia.guo@intel.com>
Acked-by: Alvin Zhang <alvinx.zhang@intel.com>

ethdev: fix build warning on 64-bit value

Building OVS with dpdk, sparse complains about 64-bit constant being
passed as a normal integer that can't fit it:
error: constant 0xffffffffffffffff is so big it is unsigned long

Fixes: ecbc8570131d ("ethdev: add PFCP header to flow API")

Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>

eal: fix typo in endian conversion macros

Caught by code inspection, for little endian, RTE_LEXX macros should
provide rte_leXX_t type values.

Fixes: b75667ef9f7e ("eal: add static endianness conversion macros")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>

ring: fix build with -Wswitch-enum

Some popular vswitch implementation might use a gcc option that
complains about missing enums in switch statements.
Fix this by listing all possible values.

Fixes: 664ff4b1729b ("ring: introduce peek style API")

Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

maintainers: update for AMD axgbe and CCP crypto

Ownership change.

Signed-off-by: Amaranath Somalapuram <asomalap@amd.com>
Acked-by: Ravi Kumar <ravi1.kumar@amd.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

maintainers: update for Arm v7 and v8

Updating ARM v7 and v8 maintainer. Gavin is leaving Arm.

Signed-off-by: Ruifeng Wang <ruifeng.wang@arm.com>
Acked-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>

version: 20.05-rc1

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>

examples/l3fwd-power: implement proper shutdown

Currently, shutdown for l3fwd-power application is all over the place
and may or may not happen either in the signal handler or in the main()
function. Fix this so that the signal handler will only set the exit
variable, thereby allowing all of the loops to end properly and proceed
to deinitialize everything.

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: David Hunt <david.hunt@intel.com>
Reviewed-by: Reshma Pattan <reshma.pattan@intel.com>

examples/l3fwd-power: exit on power lib init failure

Currently, if power library initialization fails, only a log message is
displayed. This is suboptimal for a number of reasons, but the main one
is that telemetry mode does not depend on the power library and can
therefore run in environments where l3fwd-power would normally not run
correctly (such as inside a VM). This will lead to attempts to
deinitialize the power library on exit, with a subsequent forced unclean
shutdown of DPDK.

Fix this by only initializing the power library in modes that actually
need it, and change a log message to a failure to initialize.

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: David Hunt <david.hunt@intel.com>
Reviewed-by: Reshma Pattan <reshma.pattan@intel.com>

doc: fix log level example in Linux guide

Now we need to add prefix like lib. to enable the log,
also changing val 8 to "debug"" which would be more descriptive.

Fixes: ffb9fd1b0808 ("log: update legacy modules dynamic logs regex")
Cc: stable@dpdk.org
Reported-by: Haiyue Wang <haiyue.wang@intel.com>
Signed-off-by: Xiaolong Ye <xiaolong.ye@intel.com>

timer: relax barrier for status update

Volatile has no ordering semantics. The rte_timer structure defines
timer status as a volatile variable and uses the rte_r/wmb barrier
to guarantee inter-thread visibility.

This patch optimized the volatile operation with c11 atomic operations
and one-way barrier to save the performance penalty. According to the
timer_perf_autotest benchmarking results, this patch can uplift 10%~16%
timer appending performance, 3%~20% timer resetting performance and 45%
timer callbacks scheduling performance on aarch64 and no loss in
performance for x86.

Suggested-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Erik Gabriel Carrillo <erik.g.carrillo@intel.com>

eal: disable tracing on Windows

Fix build errors caused by using Unix-specific functions in common code.
Hide and disable command-line options related to tracing on Windows.

Fixes: 3d26a70ae338 ("trace: add trace configuration parameter")
Fixes: 3b155d24bdaf ("trace: hook subsystem to Linux")

Reported-by: Pallavi Kadam <pallavi.kadam@intel.com>
Suggested-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>

app/testpmd: add parsing for QinQ VLAN headers

When having QinQ VLAN headers in the packet, parse_ethernet
is capable of parsing only the first VLAN.

Add parsing for QinQ VLAN headers in the packet.

Fixes: 51f694dd40f5 ("app/testpmd: rework checksum forward engine")
Cc: stable@dpdk.org
Signed-off-by: Raslan Darawsheh <rasland@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
Acked-by: Bernard Iremonger <bernard.iremonger@intel.com>

net/bnxt: fix VNIC Rx queue count on VNIC free

bnxt_free_one_vnic and bnxt_setup_one_vnic are called on configuring
port vlan stripping. bnxt_setup_one_vnic keeps incrementing the
vnic rx_queue_cnt. Fix to reset vnic rx_queue_cnt in bnxt_free_one_vnic.

Fixes: cfadfee41ed1 ("net/bnxt: fix VLAN strip")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>

net/bnxt: fix memory leak during queue restart

During port 0 rxq 1 start ie queue start,
bnxt_free_hwrm_rx_ring() we are clearing the pointers to mbuf array.
Due to this we overwrite the queue with fresh mbuf allocations
causing previously allocated mbufs to leak.
Add a check before allocating mbuf to replenish only empty mbuf slots
in the RxQ.

Fixes: 2eb53b134aae ("net/bnxt: add initial Rx code")
Cc: stable@dpdk.org
Signed-off-by: Rahul Gupta <rahul.gupta@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>

mem: fix build

Some compilers (on RHEL7 and CentOS7) were getting this error:
error: "RTE_EXEC_ENV_FREEBSD" is not defined [-Werror=undef]

Existence of a macro must be checked with "#ifdef" or "#if defined".

Fixes: d72e4042c5eb ("mem: exclude unused memory from core dump")

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Tested-by: Raslan Darawsheh <rasland@mellanox.com>