load-acquire memory order has a barrier. Loading it inside
the loop will result in a barrier in every iteration. Hence,
load the variable once outside the loop.
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
For zero-copy S2M rings, ring->head is updated by the sender and
eth_memif_tx_zc function is called in the context of sending thread.
The loads in the sender do not need to synchronize with its own stores.
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
eth_memif_tx_zc function is called only for S2M rings. Remove
additional code for M2S rings in this function.
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
For zero-copy M2S rings, ring->head is updated by the receiver
and eth_memif_rx_zc function is called in the context of
receiving thread. The loads in the receiver do not need to
synchronize with its own stores.
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
For S2M rings, ring->head is updated by the sender and eth_memif_tx
function is called in the context of sending thread. The loads in
the sender do not need to synchronize with its own stores.
Fixes: a2aafb9aa6 ("net/memif: optimize with one-way barrier")
Cc: stable@dpdk.org
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
For M2S rings, ring->head is updated by the receiver and eth_memif_rx
function is called in the context of receiving thread. The loads in
the receiver do not need to synchronize with its own stores.
Fixes: a2aafb9aa6 ("net/memif: optimize with one-way barrier")
Cc: stable@dpdk.org
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
For M2S rings, ring->tail is updated by the sender and eth_memif_tx
function is called in the context of sending thread. The loads in
the sender do not need to synchronize with its own stores.
Fixes: a2aafb9aa6 ("net/memif: optimize with one-way barrier")
Cc: stable@dpdk.org
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
In the case of S2M queues, the receiver synchronizes with the sender
(i.e. informs of the packets it has received) using ring->tail.
Hence, the sender does not need to update last_tail.
In the case of M2S queues, the receiver uses last_tail to
keep track of the descriptors it has received. The
sender is not required to update the last_tail. Updating
the last_tail makes it a shared variable between the
transmitter and receiver affecting the performance.
Fixes: 09c7e63a71 ("net/memif: introduce memory interface PMD")
Cc: stable@dpdk.org
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Jakub Grajciar <jgrajcia@cisco.com>
When input only Rx interface argument 'rx_iface_in' provided, current
code assigns 'eth_null_rx' burst function by mistake and no packet
received as a result.
Like in following usage no packets received from physical interface:
"--vdev net_pcap0,rx_iface_in=eth0,tx_iface=eth0"
Fixing the burst function assignment when 'rx_iface_in' argument is used
Fixes: f14a945915 ("net/pcap: remove Rx queue argument necessity")
Cc: stable@dpdk.org
Reported-by: Muthurajan Jayakumar <muthurajan.jayakumar@intel.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
In 'bond_ethdev_rx_burst_alb()' in Rx path, 'bond_rx_queue' should be
used, not 'bond_tx_queue'.
Fixes: 06fe78b98c ("bond: add mode 6")
Cc: stable@dpdk.org
Signed-off-by: Dongsheng Rong <rongdongsheng@baidu.com>
Signed-off-by: RongQing Li <lirongqing@baidu.com>
Reviewed-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Current Rx round robin policy for the slaves has two issue:
1. active_slave in bond_dev_private is shared by multiple PMDS which
maybe cause some slave Rx hungry, for example, there is two PMD and
two slave port, both PMDs start to receive, and see that active_slave
is 0, and receive from slave 0, after complete, they increase
active_slave by one, totally active_slave are increased by two, next
time, they will start to receive from slave 0 again, at last, slave 1
maybe drop packets during to not be polled by PMD
2. active_slave is shared and written by multiple PMD in RX path for
every time RX, this is a kind of cache false share, low performance.
So move active_slave from bond_dev_private to bond_rx_queue make it as
per queue variable
Fixes: ae2a04864a ("net/bonding: reduce slave starvation on Rx poll")
Cc: stable@dpdk.org
Signed-off-by: RongQing Li <lirongqing@baidu.com>
Signed-off-by: Dongsheng Rong <rongdongsheng@baidu.com>
Reviewed-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
In hn_dev_tx_queue_setup() allocated memory for txq, we don't free it
when error happens and it will lead to memory leak.
We can check for tx_free_thresh at the beginning of the function to
fix it, before calling txq = rte_zmalloc_socket().
Fixes: cc02518132 ("net/netvsc: split send buffers from Tx descriptors")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Acked-by: Long Li <longli@microsoft.com>
Some timesync related source files are built only when
'RTE_LIBRTE_IEEE1588' config is set, which is missing in meson.
Building with '-DRTE_LIBRTE_IEEE1588' cflag will enable timesync
functionality in the files that are build, but it won't build files
that require 'RTE_LIBRTE_IEEE1588' config.
This causes the build error.
Fixing by removing config check in meson file and build all .c files by
default, but wrap relevant code part with 'RTE_LIBRTE_IEEE1588' macro.
Also removing 'RTE_LIBRTE_IEEE1588' ifdef around some fields of the data
structures, to not require finer grained macro wraps on the functions
using them.
Since the registration of the function disabled with macro check, having
functions compiled shouldn't affect the functionality.
Fixes: 184c39d165 ("net/dpaa2: add DPRTC sub-module")
Cc: stable@dpdk.org
Signed-off-by: Chenxu Di <chenxux.di@intel.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Sachin Saxena <sachin.saxena@oss.nxp.com>
When reading bus and segment values using SPDRP_BUSNUMBER
bits 24-31 are undefined.
They are cleared to verify we read the segment number correctly.
Fixes: c3adf8144a ("bus/pci: support segment as address domain on Windows")
Cc: stable@dpdk.org
Signed-off-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Ranjit Menon <ranjit.menon@intel.com>
Acked-by: Narcisa Vasile <navasile@linux.microsoft.com>
This commit implements the eventdev ABI changes required by
the DLB/DLB2 PMDs. Several data structures and constants are modified
or added in this patch, thereby requiring modifications to the
dependent apps and examples.
The DLB/DLB2 hardware does not conform exactly to the eventdev interface.
1) It has a limit on the number of queues that may be linked to a port.
2) Some ports a further restricted to a maximum of 1 linked queue.
3) DLB does not have the ability to carry the flow_id as part
of the event (QE) payload. Note that the DLB2 hardware is capable of
carrying the flow_id.
Following is a detailed description of the changes that have been made.
1) Add new fields to the rte_event_dev_info struct. These fields allow
the device to advertise its capabilities so that applications can take
the appropriate actions based on those capabilities.
struct rte_event_dev_info {
uint32_t max_event_port_links;
/**< Maximum number of queues that can be linked to a single event
* port by this device.
*/
uint8_t max_single_link_event_port_queue_pairs;
/**< Maximum number of event ports and queues that are optimized for
* (and only capable of) single-link configurations supported by this
* device. These ports and queues are not accounted for in
* max_event_ports or max_event_queues.
*/
}
2) Add a new field to the rte_event_dev_config struct. This field allows
the application to specify how many of its ports are limited to a single
link, or will be used in single link mode.
/** Event device configuration structure */
struct rte_event_dev_config {
uint8_t nb_single_link_event_port_queues;
/**< Number of event ports and queues that will be singly-linked to
* each other. These are a subset of the overall event ports and
* queues; this value cannot exceed *nb_event_ports* or
* *nb_event_queues*. If the device has ports and queues that are
* optimized for single-link usage, this field is a hint for how many
* to allocate; otherwise, regular event ports and queues can be used.
*/
}
3) Replace the dedicated implicit_release_disabled field with a bit field
of explicit port capabilities. The implicit_release_disable functionality
is assigned to one bit, and a port-is-single-link-only attribute is
assigned to other, with the remaining bits available for future assignment.
* Event port configuration bitmap flags */
#define RTE_EVENT_PORT_CFG_DISABLE_IMPL_REL (1ULL << 0)
/**< Configure the port not to release outstanding events in
* rte_event_dev_dequeue_burst(). If set, all events received through
* the port must be explicitly released with RTE_EVENT_OP_RELEASE or
* RTE_EVENT_OP_FORWARD. Must be unset if the device is not
* RTE_EVENT_DEV_CAP_IMPLICIT_RELEASE_DISABLE capable.
*/
#define RTE_EVENT_PORT_CFG_SINGLE_LINK (1ULL << 1)
/**< This event port links only to a single event queue.
*
* @see rte_event_port_setup(), rte_event_port_link()
*/
#define RTE_EVENT_PORT_ATTR_IMPLICIT_RELEASE_DISABLE 3
/**
* The implicit release disable attribute of the port
*/
struct rte_event_port_conf {
uint32_t event_port_cfg;
/**< Port cfg flags(EVENT_PORT_CFG_) */
}
This patch also removes the depreciation notice and announce
the new eventdev ABI changes in release note.
Signed-off-by: Timothy McDaniel <timothy.mcdaniel@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Add minimum burst throughout the scheduler pipeline and a flush counter.
Use a single threaded ring implementation for the reorder buffer free list.
Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
Issue has been observed in case of multi segments where mbuf
data gets corrupted due to missing barriers. Changes made to
mbuf just before LMTST by one core gets updatded when the
same mbuf is in use by another core, leading to corruption.
It should be ensured that all changes made to mbuf should be
written before LMTST.
Fixes: cbd5710db4 ("net/octeontx2: add Tx multi segment version")
Cc: stable@dpdk.org
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Harman Kalra <hkalra@marvell.com>
Improve single flow performance by moving the point of coherence
to the end of transmit sequence.
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Add SWTAG flush operation at the end of transmit sequence to
immediately release the tag held by the core.
Reuse Tag address to check SWTAG completion status.
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
In the op new mode of crypto adapter, the completed crypto operation
is submitted to the event device by the OCTEON TX2 crypto PMD.
During event device dequeue the result of crypto operation is checked.
Signed-off-by: Ankur Dwivedi <adwivedi@marvell.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
The crypto adapter callback functions and associated data structures
are added.
Signed-off-by: Ankur Dwivedi <adwivedi@marvell.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Some functions are common across cryptodev pmd and the event
crypto adapter. This patch moves them into a helper file.
Signed-off-by: Ankur Dwivedi <adwivedi@marvell.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Coverity flags that 'portal' variable is used before
it's checked for NULL. This patch fixes this issue.
Coverity issue: 323516
Fixes: 4ab57b042e ("event/dpaa2: affine portal at runtime during I/O")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Acked-by: Nipun Gupta <nipun.gupta@nxp.com>
Validate events configured in ssopf against the total number of
events configured across all the RX/TIM event adapters.
Events available to ssopf can be reconfigured by passing the required
amount to kernel bootargs and are only limited by DRAM size.
Example:
ssopf.max_events= 2097152
Cc: stable@dpdk.org
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Reduce number of source BDs to submit a request to crypto engine.
This improves the performance as crypto engine fetches all the BDs in
single cycle. Adjust optional metadata (OMD) in continuation of
fixed meta data (FMD).
Signed-off-by: Vikas Gupta <vikas.gupta@broadcom.com>
Signed-off-by: Raveendra Padasalagi <raveendra.padasalagi@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Based on the new field in PDCP xform, a decision is made
to create a PDCP session with or without SDAP enabled.
Signed-off-by: Akhil Goyal <akhil.goyal@nxp.com>
Based on the new field in PDCP xform, a decision is made
to create a PDCP session with or without SDAP enabled.
Signed-off-by: Akhil Goyal <akhil.goyal@nxp.com>
The SDAP is a protocol in the LTE stack on top of PDCP. It is
dedicated to QoS.
The difficulty of implementing this protocol is because the
PDCP change behavior regarding encryption and authentication
of the SDU it receives. In effect PDCP shall not encrypt the
SDAP SDU but must authenticate it (when encryption and
authentication is enabled).
The current version of SEC does not support the SDAP and the
change of behavior of PDCP prevent the use of the PDCP
protocol command available.
The way to do it is to reuse the PDCP implementation but to
not use the PDCP protocol and to have descriptors which
performs the PDCP protocol.
It is implemented by doing small changes of code:
#ifdef SDAP_SUPPORT
length += SDAP_BYTE_SIZE;
offset -= SDAP_BYTE_SIZE;
#endif
after having computed the size of the SN to read from the
input data, then
#ifdef SDAP_SUPPORT
MATHI(p, MATH0, LSHIFT, 8, MATH1, 8, 0);
MATHB(p, MATH1, AND, sn_mask, MATH1, 8, IFB | IMMED2);
#else
MATHB(p, MATH0, AND, sn_mask, MATH1, 8, IFB | IMMED2);
#endif
It will keep the SN and the SDAP header in MATH0, then shift
it to remove the SDAP header and store the result in MATH1.
Signed-off-by: Franck Lenormand <franck.lenormand@nxp.com>
Signed-off-by: Akhil Goyal <akhil.goyal@nxp.com>
This patch updates QAT PMD to add raw data-path API support.
Signed-off-by: Fan Zhang <roy.fan.zhang@intel.com>
Acked-by: Adam Dybkowski <adamx.dybkowski@intel.com>
This patch updates ``rte_crypto_sym_vec`` structure to add
support for both cpu_crypto synchronous operation and
asynchronous raw data-path APIs. The patch also includes
AESNI-MB and AESNI-GCM PMD changes, unit test changes and
documentation updates.
Signed-off-by: Fan Zhang <roy.fan.zhang@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Acked-by: Akhil Goyal <akhil.goyal@nxp.com>
GCM digest sizes should be between 1 and 16 bytes.
Fixes: 7b2d4706c9 ("crypto/aesni_mb: support newer library version only")
Cc: stable@dpdk.org
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Digest size for CCM was being checked for other algorithms
apart from CCM.
Fixes: c4c0c312a8 ("crypto/aesni_mb: check for invalid digest size")
Cc: stable@dpdk.org
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Add Scatter-gather list support for AES-GMAC.
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Tested-by: Fan Zhang <roy.fan.zhang@intel.com>
Acked-by: Fan Zhang <roy.fan.zhang@intel.com>
Remove enumerators RTE_CRYPTO_CIPHER_LIST_END,
RTE_CRYPTO_AUTH_LIST_END, RTE_CRYPTO_AEAD_LIST_END to prevent
ABI breakages that may arise when adding new crypto algorithms.
Signed-off-by: Michael Shamis <michaelsh@marvell.com>
Removed references to RTE_CRYPTO_CIPHER_LIST_END and
RTE_CRYPTO_AUTH_LIST_END to prevent ABI breakages
that may arise when adding new crypto algorithms.
Signed-off-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
When destroying a security session, the AESNI-MB PMD attempted to clear
the private aesni_mb session object to remove any key material. However,
the function aesni_mb_pmd_sec_sess_destroy() cleared the security session
object instead of the private session object.
This patch fixes this issue by now clearing the private session object.
Fixes: fda5216fba ("crypto/aesni_mb: support DOCSIS protocol")
Cc: stable@dpdk.org
Signed-off-by: David Coyle <david.coyle@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
In case of LX2160 or SEC ERA 10, share wait has performance
optimizations wrt to ok-to-share signal which allows multiple
DECOs to work together even in case of single queue and single SA.
Hence updated the descriptor sharing only in case of ERA10.
Signed-off-by: Akhil Goyal <akhil.goyal@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
In case of LX2160 or SEC ERA >= 10, max anti replay window
size supported is 1024. For all other versions of SEC, the
maximum value is capped at 128 even if application gives
more than that.
Signed-off-by: Akhil Goyal <akhil.goyal@nxp.com>
Signed-off-by: Yi Liu <yi.liu@nxp.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
added support for non-HMAC for auth algorithms
(SHA1, SHA2, MD5).
Corresponding capabilities are enabled so that test
application can enable those test cases.
Signed-off-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Signed-off-by: Akhil Goyal <akhil.goyal@nxp.com>
DPAA2 crypto object access need availability of MCP object
pointer. In case of secondary process, we need to use local
MCP pointer instead of primary process.
Signed-off-by: Nipun Gupta <nipun.gupta@nxp.com>
Signed-off-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Acked-by: Akhil Goyal <akhil.goyal@nxp.com>
dpdk-procinfo calls the crypto stats API, which results segmentation
fault on DPAA2_SEC.
The queue pair array will be NULL, when it is used without
configuring the SEC device.
Fixes: 02f35eee26 ("crypto/dpaa2_sec: support statistics")
Cc: stable@dpdk.org
Signed-off-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Acked-by: Akhil Goyal <akhil.goyal@nxp.com>
The function handling anti replay is added. If replay window
is enabled the rx packets will be validated against the window. The
rx offload fails in case of error.
Signed-off-by: Ankur Dwivedi <adwivedi@marvell.com>
Acked-by: Akhil Goyal <akhil.goyal@nxp.com>
Initialize the inbound session for anti replay. The replay
window is allocated during session create and freed in session destroy.
Signed-off-by: Ankur Dwivedi <adwivedi@marvell.com>
Acked-by: Akhil Goyal <akhil.goyal@nxp.com>
Add crypto h/w module to process crypto op. Crypto op is processed via
sym_engine module before submitting the crypto request to HW queues.
Signed-off-by: Vikas Gupta <vikas.gupta@broadcom.com>
Signed-off-by: Raveendra Padasalagi <raveendra.padasalagi@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Acked-by: Akhil Goyal <akhil.goyal@nxp.com>
