net/ixgbe: fix crash on detach

When detaching a port bound to ixgbe PMD, if the port
does not have any VFs, *vfinfo is not set and there is
a NULL dereference attempt, when calling
rte_eth_switch_domain_free(), which expects VFs to be used,
causing a segmentation fault.

Steps to reproduce:

./testpmd -- -i
testpmd> port stop all
testpmd> port close all
testpmd> port detach 0

Bugzilla ID: 57
Fixes: cf80ba6e2038 ("net/ixgbe: add support for representor ports")
Cc: stable@dpdk.org
Reported-by: Anatoly Burakov <anatoly.burakov@intel.com>
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Tested-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Remy Horton <remy.horton@intel.com>

net/mlx5: fix error number handling

rte_errno should be saved only if error has occurred because rte_errno
could have garbage value.

Fixes: a6d83b6a9209 ("net/mlx5: standardize on negative errno values")
Cc: stable@dpdk.org
Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: clean-up developer logs

Split maintainers logs from user logs.

A lot of debug logs are present providing internal information on how
the PMD works to users.  Such logs should not be available for them and
thus should remain available only when the PMD is compiled in debug
mode.

This commits removes some useless debug logs, move the Maintainers ones
under DEBUG and also move dump into debug mode only.

Cc: stable@dpdk.org
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/ena: enable write combining

Write combining (WC) increases NIC performance by making better
utilization of PCI bus. ENA PMD may make usage of this feature.

To enable it load igb_uio driver with wc_activate set to 1.

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

bus/pci: enable write combining during mapping

Write combining (WC) increases NIC performance by making better
utilization of PCI bus, but cannot be used by all PMDs.

It will be enabled only if RTE_PCI_DRV_WC_ACTIVATE will be set in
drivers flags. For proper work also igb_uio driver must be loaded with
wc_activate set to 1.

When mapping PCI resources, firstly check if it support WC
and then try to use it.
In case of failure, it will fallback to normal mode.

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>

bus/pci: reference driver structure before mapping

Add pointer to driver structure before calling rte_pci_map_device.
It allows to use driver flags for adjusting configuration.

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

igb_uio: add write combining option

Write combining (WC) increases NIC performance by making better
utilization of PCI bus, but cannot be use by all PMD.

To get internal_addr memory need to be mapped. But as memory could not be
mapped twice: with and without WC it should be skipped for WC. [1]

To do not spoil other drivers that potentially could use internal_addr,
parameter wc_activate adds possibility to skip it for those PMDs,
that do not use it.

[1] https://www.kernel.org/doc/ols/2008/ols2008v2-pages-135-144.pdf
section 5.3 and 5.4

Signed-off-by: Rafal Kozik <rk@semihalf.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

maintainers: update for ethdev

Ferruh and Andrew are doing excellent reviews and contributions
to ethdev API.
They become official maintainers and responsibles of this major lib.

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

ipc: fix locking while sending messages

Previously, we were putting an exclusive lock to prevent secondary
processes spinning up while we are sending our messages. However,
using exclusive locks had an effect of disallowing multiple
simultaenous unrelated messages/requests being sent, which was
not the intention behind locking.

Fix it to put a shared lock on the directory. That way, we still
prevent secondary process initializations while sending data over
IPC, but allow multiple unrelated transmissions to proceed.

Fixes: 89f1fe7e6d95 ("eal: lock IPC directory on init and send")
Cc: stable@dpdk.org
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: Qi Zhang <qi.z.zhang@intel.com>

cryptodev: fix ABI breakage

In 17.08, the crypto operation was restructured,
and some reserved bytes (5) were added  to have the mempool
pointer aligned to 64 bits, since the structure is expected
to be aligned to 64 bits, allowing future additions with no
ABI breakage needed.

In 18.05, a new 2-byte field was added, so the reserved bytes
were reduced to 3. However, this field was added after the first 3 bytes
of the structure, causing it to be placed in an offset of 4 bytes,
and therefore, forcing the mempool pointer to be placed after 16 bytes,
instead of a 8 bytes, causing unintentionally the ABI breakage.

This commit fixes the breakage, by swapping the reserved bytes
and the private_data_offset field, so the latter is aligned to 2 bytes
and the offset of the mempool pointer returns to its original offset,
8 bytes.

Fixes: 54c836846603 ("cryptodev: set private data for session-less mode")
Cc: stable@dpdk.org
Reported-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Acked-by: Abhinandan Gujjar <abhinandan.gujjar@intel.com>

cryptodev: convert to SPDX license tag

Converted the license header of the files
that still have the old full header.

Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: Kevin Laatz <kevin.laatz@intel.com>

doc: document QAT debug options

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move to dynamic logging for data-path trace

For all trace on the data-path move to dynamic logging.
Add new QAT_DP_LOG macro to separate from non-dp trace.
Use the overall RTE_LOG_DP_LEVEL to override and compile out
debug data-path trace.
Add "unlikely" in two appropriate error checks.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: rename log macro for non-dp logs

Rename PMD_DRV_LOG to the more appropriate name QAT_LOG.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move to dynamic logging for non-dp trace

For all trace not on the data-path move to dynamic logging.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: clean up unused and useless trace

Removed INIT_FUNC trace and other unused macros.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: remove configurable max number of sessions

This patch removes CONFIG_RTE_QAT_PMD_MAX_NB_SESSIONS
from common_base, config/rte_config.h files and
defines QAT_SYM_PMD_MAX_NB_SESSIONS inside qat_sym_pmd.h
file instead.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Acked-by: Fiona Trahe <fiona.trahe@intel.com>

doc: specify QAT driver and device name formats

Document the driver and device naming formats.
Changed the underscores alignment.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: check for service type

Other services, apart from symmetric crypto,
such as compression, will be added in future patches.
Therefore, the assumption that only symmetric crypto
operations are processed will not be valid anymore,
and service type needs to be checked.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: make response process function inline

Optimize the dequeue function by inlining the response
processing function, assuming only symmetric
operations are supported.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: remove unused arguments

Process response function is only implemented
for crypto symmetric operations, which do not require
some of the arguments.
Therefore, these arguments can be removed from the
function prototype.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: optimize ADF modulo function

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: add max PCI devices to config file

Added CONFIG_RTE_PMD_QAT_MAX_PCI_DEVICES to
build config files.

Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: register appropriately named device

For every QAT PCI device probed, populate a local rte_device
containing an rte_driver. The rte_driver was created in a previous
patch to provide a crypto-specific driver name: "crypto_qat".
This was previously only used for driver registration, now it's also
used in device creation.
This allows applications to find devices driven by "crypto_qat".

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move macro to common file

QAT_64_BTYE_ALIGN_MASK macro is not specific to
symmetric crypto, but it is common for other
future services.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: remove unused macro

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: free cookie pool on queue creation error

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: modify debug message

Removed crypto reference in common debug message.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: rename variables

Renamed sgl_cookie to cookie and qp_gen_config
to qat_gen_config, as it is intended to hold
more than just queue pair data.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: remove incorrect usage of bundle number

As bundle_num is included in qat_gen1_qps static array
there shouldn't be a multiplier used in qat_qps_per_service()
Then removed ADF_NUM_BUNDLES_PER_DEV as no longer used.
Also renamed ADF_MAX_QPS_PER_BUNDLE to ADF_MAX_QPS_ON_ANY_SERVICE
and reduced from 4 to 2 which is enough for all current devices.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: add lock around CSR access and change logic

Add lock around accesses to the arbiter CSR
and use & instead of ^ as ^ not safe if
arb_disable called when already disabled.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move code into appropriate files

Move all code into appropriate files, no actual code changes. Specifically:
 - Rename rte_qat_cryptodev.c to qat_sym_pmd.c
 - Create qat_sym_pmd.h and populate with fn prototypes for qat_sym_pmd.c
 - Create qat_comp_pmd.c/.h and populate with placeholder functions
 - Create qat_asym_pmd.c/.h and populate with placeholder functions
 - Rename qat_crypto_capabilities.h to qat_sym_capabilities.h
 - Move CRYPTODEV_NAME_QAT_SYM_PMD from qat_common.h to qat_sym_pmd.h
 - Move qat_sym_dev_private from qat_device.h to qat_sym_pmd.h
 - Move prototype for qat_sym_dev_info_get frm qat_device.h 2 qat_sym_pmd.h
 - Move all qat_device.c sym dev_ops fns to qat_sym_pmd.c file
 - Move all qat_sym.c dev_ops fns to qat_sym_pmd.c file
 - Remove unused header file #includes from all files.
 - Move pci_id_qat_map, probe/release/register from
   rte_qat_cryptodev.c to qat_device.c
 - Moved stray comment for bpi_cipher_ctx_init() from qat_sym.c
   to qat_sym_session.c
 - Changed all files to use SPDX license header

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: rename functions which depend on cryptodev

Rename all device functions which depend on cryptodev structs.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: use common stats structures

Split qat_sym_stats_get/reset into 2 functions, a wrapper function calling
a new qat_stats_get/reset function which can be called per service.
Remove cryptodev stats struct from qat_qp, replace with qat_common_stats.
Add links for qat_qp into qat_pci_device using an array per service to
avoid need for a lock and so qp_id for the service can be used as index.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move to using new device structure

Struct qat_pmd_private held the data needed by cryptodev, common code
now gets most data from struct qat_pci_device instead. qat_pmd_private
is trimmed to hold only sym crypto data and renamed qat_sym_private
to reflect its usage.
Specifically
 - remove max_nb_queue_pairs from qat_pmd_private, get from qp_hw_data
 - remove max_nb_sesssions from qat_pmd_private as not needed.
 - remove qat_gen from qat_pmd_private, get from qat_pci_device instead.
 - use qat_pci_device throughout common code instead of qat_pmd_private
 - rename qat_pmd_private to qat_sym_dev_private - this now holds only
   sym-specific data for the cryptodev API
 - extend pci device name to <bdf>_qat for clarity, was just <bdf>
 - update qp mem and cookiepool names to reflect the appropriate device,
   service and qp.
 - rename qat_dev_info_get() to qat_sym_dev_info_get() as mostly sym,
   not enough common info to warrant a generic fn.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: use generic driver name for PCI registration

The QAT PMD used to register with PCI using the name "crypto_qat".
Keep this name for the driver registered with cryptodev
and use a more generic name "qat" for the PCI registration.
This paves the way for the PCI device to host other services.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: add PCI device struct

- Added struct qat_pci_device to use internally in QAT PMD
   to avoid dependencies on rte_cryptodev or rte_compressdev
 - Added a global array of these
 - Restructured probe/release to separate QAT common init/clear
   from crypto pmd create/destroy.
 - In QAT common part allocated a qat_pci_device and populated it
 - Removed meaningless check in probe for RTE_PROC_PRIMARY

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move sgl related element to appropriate files

Move SGL (Scatter-Gather List) related functions to common file
Move qat_sym_op_cookie struct to sym header file

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: rename sgl related objects

Change SGL (Scatter-Gather List) related structs and member names

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>

crypto/qat: create structures to support various generations

Create data structures to support different generations
of qat hardware supplying services through different queue pairs.
 - Add two new structs qat_gen_hw_data and qat_qp_hw_dat
 - Add a qat_service_type enum
An array of qat_qp_hw_data elements is
initialised with constants, these are arranged so that the qp_id used
on the API can be used as an index to pick up the qp data to use.
The constants are common to current generations,
new arrays will be added for future generations.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move defines from sym to qp header file

Move defines related to coalescing from sym header file to qp header
file as these will be common for all services.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: remove dependencies on cryptodev from common

Remove dependence on rte_cryptodev from common qp code
to facilitate being used by other device types in future.
Transferred required data into qat-specific structures.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move sym-specific qp code to sym file

Move sym qp setup code from qat_qp.c to qat_sym.c

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: separate sym-specific from generic qp setup

Extracted all sym-specific code from qp setup fns, leaving
generic qat_qp_setup fn and helper fns. Created a new
meta-data struct qat_qp_config to hold all the data needed
to create a qp, filled this out in the sym-specific code
and passed to the generic qp_setup fn.
No need now for rx and tx queue_create fns, one generic
queue_create fn replaces these.
Included the service name (e.g. "sym") in the qp memzone
and cookie pool names.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: move generic qp function to qp file

Move the generic enqueue and dequeue fns from
the qat_sym.c file to the qat_qp.c file
Move generic qp structs to a new qat_qp.h file

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: make dequeue function generic

Queue-handling code in dequeue is made generic, so it can
be used by other services in future. This is done by
 - Removing all sym-specific refs in input params - replace with void ptrs.
 - Wrapping this generic dequeue with the sym-specific dequeue
   called through the API.
 - extracting the sym-specific response processing into a new fn.
 - Setting a fn ptr for process_response in qp on qp creation
 - Passing void * params to this, in the service-specific implementation
   qat_sym_process_response cast back to sym structs.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: make enqueue function generic

Queue-handling code in enqueue is made generic, so it can
be used by other services in future. This is done by
 - Removing all sym-specific refs in input params - replace with void ptrs.
 - Wrapping this generic enqueue with the sym-specific enqueue
   called through the API.
 - Setting a fn ptr for build_request in qp on qp creation
 - Passing void * params to this, in the service-specific implementation
   qat_sym_build_request cast back to sym structs.

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: rename sym-specific structs

qat_session           -> qat_sym_session
qat_crypto_proto_flag -> qat_sym_proto_flag
qat_alg_cd            -> qat_sym_cd
qat_crypto_op_cookie  -> qat_sym_op_cookie
qat_gen1_capabilities -> qat_gen1_sym_capabilities
qat_gen2_capabilities -> qat_gen2_sym_capabilities

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: rename functions for consistency

Rename fn names to shorten them, i.e.
  qat_crypto_sym_xxx to qat_sym_xxx
  _content_desc_ to _cd_
Renaming symmetric crypto specific with consistent names:
  qat_crypto_set_session_parameters->qat_sym_set_session_parameters
  qat_write_hw_desc_entry()->qat_sym_build_request()
  qat_alg_xxx ->qat_sym_xxx
  qat_sym_xxx_session_yyy()->qat_sym_session_xxx_yyy()
Removed unused prototypes:
  qat_get_inter_state_size()
  qat_pmd_session_mempool_create()
Removed 2 unnecessary extern declarations

Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: change filename crypto to sym

This commit renames qat_crypto.c/h to qat_sym.c/h
And makes a few whitespace changes to resolve line-length
issues.

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: add symmetric session file

This commit adds qat_sym_session.c/h files and moves objects
from qat_algs_build_desc and qat_algs.h

Following objects were moved:
qat_adf/qat_algs_build_desc.c => qat_sym_session.c
- all objects -
qat_adf/qat_algs.h => qat_sym_session.h
- enum qat_crypto_proto_flag
- struct qat_alg_cd
- struct qat_session
- int qat_get_inter_state_size()
- int qat_alg_aead_session_create_content_desc_cipher()
- int qat_alg_aead_session_create_content_desc_auth()
- void qat_alg_init_common_hdr()
- int qat_alg_validate_aes_key()
- int qat_alg_validate_aes_docsisbpi_key()
- int qat_alg_validate_snow3g_key()
- int qat_alg_validate_kasumi_key()
- int qat_alg_validate_3des_key()
- int qat_alg_validate_des_key()
- int qat_cipher_get_block_size()
- int qat_alg_validate_zuc_key()
-- all macros
qat_crypto.h => qat_sym_session.h
int qat_crypto_sym_configure_session()
int qat_crypto_set_session_parameters()
int qat_crypto_sym_configure_session_aead()
int qat_crypto_sym_configure_session_cipher()
int qat_crypto_sym_configure_session_auth()
int qat_alg_aead_session_create_content_desc_cipher()
int qat_alg_aead_session_create_content_desc_auth()
static struct rte_crypto_auth_xform qat_get_auth_xform()
static struct rte_crypto_cipher_xform qat_get_cipher_xform()

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: remove unused includes

This commit removes unused includes from qat_crypto.c

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>

crypto/qat: add device files

This commit adds new qat_device file.
Following objects were moved:

qat_crypto.h => qat_device.h
- struct qat_pmd_private
- uint8_t cryptodev_qat_driver_id
- int qat_crypto_sym_qp_release (EXTERN)
- int qat_dev_config()
- int qat_dev_start()
- void qat_dev_stop()
- int qat_dev_close()
- void qat_dev_info_get()
qat_crypto.c => qat_device.c
- int qat_dev_config()
- int qat_dev_start()
- void qat_dev_stop()
- int qat_dev_close()
- void qat_dev_info_get()

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

crypto/qat: add common header

This commit adds qat_common.h header file.
Following objects were moved to it:

qat_algs.h =>. qat_common.h
- struct qat_alg_buf
- struct qat_alg_buf_list
- struct qat_crypto_op_cookie
- QAT_SGL_MAX_NUMBER
qat_crypto.h => qat_common.h
- CRYPTODEV_NAME_QAT_SYM_PMD

Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Signed-off-by: Fiona Trahe <fiona.trahe@intel.com>

net: rename u16 to fix shadowed declaration

This patch renames u16 to u16_buf. u16 as a variable name causes a shadowed
declaration warning if, for example, the application also typedefs u16
(e.g. by including a header containing "typedef unsigned short u16") and
the application is built with -Wshadow.

Signed-off-by: Gage Eads <gage.eads@intel.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>

bus/pci: remove unused function to detach by address

The function rte_pci_detach() is private to PCI and is
not used anywhere in current code base. Remove dead code.

Signed-off-by: Rami Rosen <rami.rosen@intel.com>

bus/pci: remove unused function to probe by address

The function rte_pci_probe_one is private to PCI and is
not used anywhere in current code base. Remove dead code.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

bus/pci: remove unused unbind function prototype

Prototype for pci_unbind_kernel_driver exists but no code.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Gaetan Rivet <gaetan.rivet@6wind.com>

bus/pci: make remove function static

Only used in one file, and therefore can be made static.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Gaetan Rivet <gaetan.rivet@6wind.com>

cmdline: remove useless intermediate buffer

Rather than copy the string, we can use a precision in the format string
given to printf.

Signed-off-by: David Marchand <david.marchand@6wind.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>

log: remove useless intermediate buffer

Rather than copy the log message, we can use a precision in the format
string given to syslog.

Signed-off-by: David Marchand <david.marchand@6wind.com>
Reviewed-by: Olivier Matz <olivier.matz@6wind.com>

igb_uio: fail and log if kernel lock down is enabled

When EFI secure boot is enabled, it is possible to lock down kernel and
prevent accessing device BARs and this makes igb_uio unusable.

Lock down patches are not part of the vanilla kernel but they are
applied and used by some distros already [1].

It is not possible to fix this issue, but intention of this patch is to
detect and log if kernel lock down enabled and don't insert the module
for that case.

The challenge is since this feature enabled by distros, they have
different config options and APIs for it. This patch is done based on
Fedora and Ubuntu kernel source, may needs to add more distro specific
support.

[1]
kernel.ubuntu.com/git/ubuntu/ubuntu-artful.git/commit/?id=99f9ef18d5b6
And a few more patches too.

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Luca Boccassi <bluca@debian.org>

kni: fix build with gcc 8.1

Error observed when CONFIG_RTE_KNI_KMOD_ETHTOOL config option is
enabled.

build error:
In function ‘strncpy’,
    inlined from ‘igb_get_drvinfo’ at
    kernel/linux/kni/igb_ethtool.c:814:2:
    include/linux/string.h:246:9: error: ‘__builtin_strncpy’ output
    may be truncated copying 31 bytes from a string of length 42
    [-Werror=stringop-truncation]
      return __builtin_strncpy(p, q, size);
               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Fixed by using strlcpy instead of strncpy.

adapter->fw_version size kept same because of
c3698192940c ("kni: fix build with gcc 7.1")

Also next line strncpy usage replaced with strlcpy while arround.

Fixes: c3698192940c ("kni: fix build with gcc 7.1")
Cc: stable@dpdk.org
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>

maintainers: update for tep_termination example

Xiaoyun Li has agreed to take over the maintainership of example
application tep_termination, as Jijiang Liu is no longer working
on that.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

maintainers: update Intel sub-tree committers

For subtree of dpdk-next-net-intel, Qi Zhang has agreed to take
the committer role, to replace Helin Zhang. Also Beilei Xing has
agreed to be the backup committer of the subtree.

Signed-off-by: Helin Zhang <helin.zhang@intel.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

bus/dpaa: fix build

The DPAA bus driver is defining some macros without prefix.
So it can conflict with other libraries like libbsd:

drivers/bus/dpaa/include/compat.h:53:
error: "__packed" redefined
/usr/include/bsd/sys/cdefs.h:120:
note: this is the location of the previous definition

Fixes: 39f373cf015a ("bus/dpaa: add compatibility and helper macros")
Cc: stable@dpdk.org
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>

net/ena: fix FreeBSD build

The FreeBSD define was removed from platform file when HAL was upgraded
to newer version.

In file included from drivers/net/ena/ena_ethdev.c:45:
from drivers/net/ena/ena_ethdev.h:42:
from drivers/net/ena/base/ena_com.h:37:
drivers/net/ena/base/ena_plat.h:48:2: error: "Invalid platform"

Fixes: 3adcba9a8987 ("net/ena: update HAL to the newer version")

Signed-off-by: Michal Krawczyk <mk@semihalf.com>

eventdev: fix Rx SW adapter stop

The Rx adapter stop call does not guarantee that the
SW service function will not execute after the
rte_event_eth_rx_adapter_stop() call.

Add a "started" flag to prevent the adapter from executing
if stop has been called.

Fixes: 9c38b704d280 ("eventdev: add eth Rx adapter implementation")
Cc: stable@dpdk.org
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com>

eventdev: fix internal port logic in Rx adapter

Set the internal_event_port flag when the ethdev-eventdev
packet transfer is implemented in hardware and add a check
for the flag to ignore the connection when setting up the
WRR polling sequence.

Fixes: 9c38b704d280 ("eventdev: add eth Rx adapter implementation")
Cc: stable@dpdk.org
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com>
Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>

eventdev: add event buffer flush in Rx adapter

Add an event buffer flush when the current invocation
of the Rx adapter is completed.

This patch provides lower latency in case there is a
BATCH_SIZE of events in the event buffer.

Cc: stable@dpdk.org
Suggested-by: Narender Vangati <narender.vangati@intel.com>
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com>
Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>

eventdev: fix missing update to Rx adaper WRR position

After dequeuing Rx packets and enqueueing them to the
temporary buffer towards eventdev, the packet Rx loop exits
if the temporary buffer is full but the current WRR position
is not saved.

Save away the current value of the WRR position, so packets
are dequeued from the correct Rx queue in the next invocation.

Fixes: 9c38b704d280 ("eventdev: add eth Rx adapter implementation")
Cc: stable@dpdk.org
Suggested-by: Gage Eads <gage.eads@intel.com>
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com>

test/eventdev: add multi-ports test

Add a new test to enhance the test suite, to allow multiple eth ports
rx queue to be added to rx bridge adapter. Update the test function to
reflect change of port index from 8 to 16 bit

Signed-off-by: Vipin Varghese <vipin.varghese@intel.com>
Acked-by: Nikhil Rao <nikhil.rao@intel.com>
Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>

doc: fix octeontx eventdev selftest argument

Fixes: 3516327e00fd ("event/octeontx: add selftest to device arguments")
Cc: stable@dpdk.org
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>

doc: improve eventdev library guide

Add small amount of additional code, use consistent variable names
across code blocks, change the image to represent queues and
CPU cores intuitively. These help improve the eventdev library
documentation.

Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>

eventdev: fix port in Rx adapter internal function

The dev_id parameter to fill_event_buffer() should be 16 bit,
also rename to to eth_dev_id to avoid confusion with event device
id elsewhere in the file.

Fixes: c2189c907dd1 ("eventdev: make ethdev port identifiers 16-bit")
Cc: stable@dpdk.org
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com>
Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>

net/mlx5: fix log initialization

The mlx5 driver had two init functions, but this could
cause log initialization to be done after the
other initialization. Also, the name of the function does
not match convention (cut/paste error?).

Fix by initializing log type first at start of the pmd_init.
This also gets rid of having two constructor functions.

Fixes: a170a30d22a8 ("net/mlx5: use dynamic logging")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: fix crash in device probe

This patch initializes counter descriptor struct before invoking Verbs
api to avoid segmentation fault.

Fixes: 9a761de8ea14 ("net/mlx5: flow counter support")
Cc: stable@dpdk.org
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: fix error message in probe function

Error values passed to strerror() must be positive.

Fixes: 012ad9944dfc ("net/mlx5: fix probe return value polarity")
Cc: stable@dpdk.org
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>

net/mlx5: fix missing errno in probe function

Fixes: b43802b4bdfe ("net/mlx5: support 16 hardware priorities")
Cc: stable@dpdk.org
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>

net/mlx5: fix errno object in probe function

Fixes: a6d83b6a9209 ("net/mlx5: standardize on negative errno values")
Cc: stable@dpdk.org
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>

net/mlx5: remove limitation on number of instances

This artificial limitation was inherited from the mlx4 code base and has no
purpose other than adding unnecessary noise.

This patch is a port of commit f2318196c71a ("net/mlx4: remove limitation
on number of instances").

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>

net/mlx4: fix minor resource leak during init

Temporary IB device context and list are not freed in case of a successful
initialization of the device.

This issue is caused by the two following commits, the first of which
causes initialization to return early, while the second one goes a bit
overboard while switching to negative errno values; an internal variable
(err) is needed to tell success from failure at the end of the function
since rte_errno is not reliable enough.

Fixes: f2318196c71a ("net/mlx4: remove limitation on number of instances")
Fixes: 9d14b27308a0 ("net/mlx4: standardize on negative errno values")
Cc: stable@dpdk.org
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

mbuf: fix typo in IPv6 macro comment

Comment should mention IPv6, not IPv4.

Fixes: f295a00a2b44 ("mbuf: add definitions of unified packet types")
Cc: stable@dpdk.org
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/bonding: fix MAC address reset

Currently when resetting bond mac address, we are getting the
persisted mac address from slave info considering primary port
as index. But we need to compare the port id from slave info
with the primary port to get the primary slave index and get
the persisted mac address. Without this fix, persisted mac addr
will be zero and rte_eth_dev_default_mac_addr_set will fail.

Fixes: a45b288ef21a ("bond: support link status polling")
Cc: stable@dpdk.org
Signed-off-by: Kiran Kumar <kiran.kokkilagadda@caviumnetworks.com>

vhost: use SMP memory barrier before kicking guest

vhost_vring_call() used rte_mb(), which translates into
mfence instruction on x86.

This patch changes to use rte_smp_mb(), which changed recently
to translate into a locked ADD instruction for performance
reason.

The measured gain is up to 3% with the testpmd benchmarks.

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>

vhost: introduce new function helper

Introduce an new common helper to avoid redundancy.

Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/ifc: enable the host notifier support

The necessary vDPA ops have already been implemented
in ifcvf driver. So just need to announce the necessary
protocol features to enable the host notifier support.

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Acked-by: Xiao Wang <xiao.w.wang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: support host notifier

When a vDPA device is attached, vhost user will try to
register host notifiers to QEMU to allow notifications
to be delivered between the driver in the guest and the
vDPA device in the host directly.

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: refine new device function

Make sure find avalid device id before allocating
virtio_net, if not, return directly. It may avoid
allocating and freeing virtio_net when there is
not valid device id.

Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/bonding: always update bonding link status

When the first ETH_LINK_UP slave is added to a bonding device, it is
immediately activated.  If these are polling slaves, there will be a
later link state event, when last_link_status doesn't match link_status.
But because this slave is already activated, the bonding device's link
status isn't updated.

To fix this, we always attempt to update the bonding device's link status
in the link state change callback.

Fixes: 46fb43683679 ("bond: add mode 4")
Cc: stable@dpdk.org
Signed-off-by: Chas Williams <chas3@att.com>

app/testpmd: fix crash when attaching a device

Below commit checks global device information to determine if a port uses
the softnic driver once initialized. Problem is that this information is
not available at this point when a port is initialized interactively
through a "port attach XXX" command, crashing testpmd.

This patch systematically initializes global device information to address
this issue.

Fixes: 5b590fbe09b6 ("app/testpmd: add traffic management forwarding mode")
Cc: stable@dpdk.org
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Bernard Iremonger <bernard.iremonger@intel.com>

net/vhost: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/szedata2: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/softnic: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/nfp: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/liquidio: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Shijith Thotton <shijith.thotton@caviumnetworks.com>

net/kni: add to meson build

Since the kni library is an optional library, we need to add a check
here to ensure that the kni wrapper PMD is only built when the main kni
lib is being built.

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/failsafe: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/ena: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/cxgbe: add to meson build

Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>