RE: [PATCH v2 06/16] net/dpaa2: support multiple txqs en-queue for ordered

> -Original Message- > From: Nipun Gupta > Sent: 03 January 2022 11:17 > To: Stephen Hemminger > Cc: dev@dpdk.org; tho...@monjalon.net; ferruh.yi...@intel.com; Hemant > Agrawal ; Jun Yang > Subject: RE: [PATCH v2 06/16] net/dpaa2: support multiple txqs en-queue for > ordered > > > >

RE: [PATCH v2] app/eventdev: add crypto producer mode

Hi Shijith, > -Original Message- > From: Shijith Thotton > Sent: Monday, January 3, 2022 11:34 AM > To: Gujjar, Abhinandan S ; dev@dpdk.org; Jerin > Jacob Kollanukkaran > Cc: Anoob Joseph ; Pavan Nikhilesh Bhagavatula > ; Akhil Goyal > Subject: RE: [PATCH v2] app/eventdev: add crypto pr

RE: [PATCH v2] app/eventdev: add crypto producer mode

>> >> >> >> In crypto producer mode, producer core enqueues cryptodev with >> >> software generated crypto ops and worker core dequeues crypto >> >> completion events from the eventdev. Event crypto metadata used for >> >> above processing is pre- populated in each crypto session. >> >> >> >> Param

[PATCH v3 00/15] features and fixes on NXP eth devices

From: Nipun Gupta This series adds few features and important fixes on DPAA, PFE and ENETC devices. Features added: - level 2 support for shaping on DPAA2 - loopback configuration for DPNI devices on DPAA2 - Multiple TXQ's enqueue for ordered queues for performance - VFs support on ENETC Fixes:

[PATCH v3 01/15] bus/fslmc: update MC to 10.29

From: Nipun Gupta update MC firmware library version to 10.29 Signed-off-by: Nipun Gupta Signed-off-by: Gagandeep Singh --- drivers/bus/fslmc/mc/fsl_dpmng.h | 2 +- drivers/net/dpaa2/mc/dpdmux.c | 8 ++ drivers/net/dpaa2/mc/dpkg.c | 7 +- drivers/net/dpaa2/mc/dpni

[PATCH v3 02/15] bus/fslmc: use dmb oshst for synchronization before I/O

From: Nipun Gupta Outer Shareable Store (oshst) is sufficient for Data Memory Barrier (dmb) when doing IO on the interface via QBMAN. This will sync L3/DDR with the L1/L2 cached data. Signed-off-by: Nipun Gupta --- drivers/bus/fslmc/qbman/include/compat.h | 4 ++-- 1 file changed, 2 insertion

[PATCH v3 03/15] net/dpaa2: warn user in case of high nb desc

From: Rohit Raj Added warning message if application is configuring nb_desc more than supported by PEB memory suggesting user to configure HW descriptors in normal memory rather than in faster PEB memory. Signed-off-by: Rohit Raj --- drivers/net/dpaa2/dpaa2_ethdev.c | 13 + 1 file

[PATCH v3 04/15] net/dpaa2: fix unregistering interrupt handler

From: Vanshika Shukla This patch fixes code that handles unregistering LSC interrupt handler in dpaa2_dev_stop API. Fixes: c5acbb5ea20e ("net/dpaa2: support link status event") Cc: sta...@dpdk.org Signed-off-by: Vanshika Shukla --- drivers/net/dpaa2/dpaa2_ethdev.c | 7 ++- 1 file changed,

[PATCH v3 05/15] net/dpaa2: fix timestamping for IEEE1588

From: Vanshika Shukla The current implementation of DPAA2 driver code is such that it records Rx and Tx timestamp for PTP without checking if they are PTP packets or not. Packets for which RTE_MBUF_F_RX_IEEE1588_TMST and RTE_MBUF_F_TX_IEEE1588_TMST is not set, Rx and Tx timestamp should not be re

[PATCH v3 06/15] net/dpaa2: support multiple txqs en-queue for ordered

From: Jun Yang Support the tx enqueue in order queue mode, where queue id for each event may be different. Signed-off-by: Jun Yang --- drivers/event/dpaa2/dpaa2_eventdev.c | 12 ++- drivers/net/dpaa2/dpaa2_ethdev.h | 4 + drivers/net/dpaa2/dpaa2_rxtx.c | 142 ++

[PATCH v3 07/15] net/dpaa2: add support for level 2 in traffic management

From: Gagandeep Singh This patch adds support for level 2 for QoS shaping. Signed-off-by: Gagandeep Singh --- doc/guides/nics/dpaa2.rst | 2 +- drivers/net/dpaa2/dpaa2_ethdev.c| 55 ++- drivers/net/dpaa2/dpaa2_ethdev.h| 6 +- drivers/net/dpaa2/dpaa2_tm.c| 563 +++

[PATCH v3 08/15] net/dpaa2: secondary process handling for dpni

From: Jun Yang This change uses 'dev->process_private' instead of 'priv->hw' to get dpmcp per process while setting flow distribution, as priv->hw is only valid for primary process. It also initialize rte_dpaa2_bpid_info in secondary process. Signed-off-by: Jun Yang --- drivers/mempool/dpaa2/d

[PATCH v3 09/15] bus/fslmc: add and scan dprc devices

From: Jun Yang In order to get connection endpoint of each objects, scan the dprc object. Signed-off-by: Jun Yang Signed-off-by: Nipun Gupta --- drivers/bus/fslmc/fslmc_bus.c| 15 ++- drivers/bus/fslmc/fslmc_vfio.c | 18 +++- drivers/bus/fslmc/mc/dprc.c |

[PATCH v3 10/15] net/dpaa2: support recycle loopback port

From: Jun Yang DPAA2 recycle port is used for configuring the device in the loopback mode. Loopback configuration can be at dpni level or at serdes level. Signed-off-by: Jun Yang --- drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 3 +- drivers/net/dpaa2/dpaa2_ethdev.c| 32 +- drivers/net

[PATCH v3 11/15] net/dpaa: check status before configuring shared MAC

From: Nipun Gupta For shared MAC interface, it is a prerequisite to enable the interface in the kernel, before using it in user-space. This patch makes sure that device is not getting configured in case shared MAC interface is not enabled in the kernel. Signed-off-by: Nipun Gupta --- drivers/b

[PATCH v3 12/15] net/dpaa: enable checksum for shared MAC interface

From: Nipun Gupta In case of shared MAC B0V bit in contextA is required to be set to set so that ASPID is 0. Signed-off-by: Brick Yang Signed-off-by: Nipun Gupta --- drivers/net/dpaa/dpaa_ethdev.c | 4 1 file changed, 4 insertions(+) diff --git a/drivers/net/dpaa/dpaa_ethdev.c b/drivers

[PATCH v3 13/15] net/enetc: add support for VFs

From: Gagandeep Singh Add virtual function support for enetc devices Signed-off-by: Gagandeep Singh --- drivers/net/enetc/enetc_ethdev.c | 25 - 1 file changed, 20 insertions(+), 5 deletions(-) diff --git a/drivers/net/enetc/enetc_ethdev.c b/drivers/net/enetc/enetc_eth

[PATCH v3 14/15] net/pfe: reduce driver initialization time

From: Gagandeep Singh This patch reduces the delay in the device init. Signed-off-by: Gagandeep Singh Signed-off-by: Nipun Gupta --- drivers/net/pfe/pfe_hif.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/pfe/pfe_hif.c b/drivers/net/pfe/pfe_hif.c index c4

[PATCH v3 15/15] net/pfe: remove setting unused value

From: Apeksha Gupta remove setting link status where it is not being used Signed-off-by: Apeksha Gupta --- drivers/net/pfe/pfe_ethdev.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/net/pfe/pfe_ethdev.c b/drivers/net/pfe/pfe_ethdev.c index 047010e15e..e5aaf5dcfd

Re: [RFC 1/1] vhost: integrate dmadev in asynchronous datapath

Hi Jiayu, On 12/28/21 02:15, Hu, Jiayu wrote: Hi Maxime, Thanks for your comments, and some replies are inline. Thanks, Jiayu -Original Message- From: Maxime Coquelin Sent: Friday, December 24, 2021 6:40 PM To: Hu, Jiayu ; dev@dpdk.org Cc: i.maxim...@ovn.org; Xia, Chenbo ; Richardso

[Bug 913] [dpdk-19.11.11]'mk' makefile build failed on Freebsd13 with clang11.0.1

https://bugs.dpdk.org/show_bug.cgi?id=913 Christian Ehrhardt (christian.ehrha...@canonical.com) changed: What|Removed |Added Resolution|--- |FIXED

[PATCH 0/1] Minor mistake in ring (en|de)queueing

Hi all, Recently I was going through the ring implementation and I believe I've found a small mistake. Not a functional error, just a slightly suboptimal behaviour for the specific case when we want to enqueue exactly the number of elements that we can before wrapping to the ring beginning (the s

[PATCH 1/1] ring: fix off by 1 mistake

When enqueueing/dequeueing to/from the ring we try to optimize by manual loop unrolling. The check for this optimization looks like: if (likely(idx + n < size)) { where 'idx' points to the first usable element (empty slot for enqueue, data for dequeue). The correct comparison here shoul

[PATCH v3] mempool: fix the description of some function return values

In rte_mempool_ring.c, the committer uses the symbol ENOBUFS to describe the return value of function common_ring_sc_dequeue, but in rte_mempool.h, the symbol ENOENT is used to describe the return value of function rte_mempool_get. If the user of dpdk uses the symbol ENOENT as the judgment conditio

Re: [PATCH v2] mempool: fix the description of some function return values

Thank you for pointing out my problem, I will fix it in the next version > 2021年12月22日 下午7:45，Morten Brørup 写道： > >> From: Zhiheng Chen [mailto:chenzhiheng0...@gmail.com >> ] >> Sent: Wednesday, 22 December 2021 09.26 >> >> Compared to patch version 1, this ve

RE: [PATCH 1/1] ring: fix off by 1 mistake

+Ring queue maintainers: Honnappa Nagarahalli , Konstantin Ananyev > From: Andrzej Ostruszka [mailto:a...@semihalf.com] > Sent: Monday, 3 January 2022 15.22 > > When enqueueing/dequeueing to/from the ring we try to optimize by > manual > loop unrolling. The check for this optimization looks li

[PATCH 0/8] ethdev: introduce IP reassembly offload

As discussed in the RFC[1] sent in 21.11, a new offload is introduced in ethdev for IP reassembly. This patchset add the RX offload and an application to test it. Currently, the offload is tested along with inline IPsec processing. It can also be updated as a standalone offload without IPsec, if t

[PATCH 1/8] ethdev: introduce IP reassembly offload

IP Reassembly is a costly operation if it is done in software. The operation becomes even more costlier if IP fragmants are encrypted. However, if it is offloaded to HW, it can considerably save application cycles. Hence, a new offload RTE_ETH_RX_OFFLOAD_IP_REASSEMBLY is introduced in ethdev for d

[PATCH 2/8] ethdev: add dev op for IP reassembly configuration

A new ethernet device op is added to give application control over the IP reassembly configuration. This operation is an optional call from the application, default values are set by PMD and exposed via rte_eth_dev_info. Application should always first retreive the capabilities from rte_eth_dev_inf

[PATCH 3/8] ethdev: add mbuf dynfield for incomplete IP reassembly

Hardware IP reassembly may be incomplete for multiple reasons like reassembly timeout reached, duplicate fragments, etc. To save application cycles to process these packets again, a new mbuf ol_flag (RTE_MBUF_F_RX_IPREASSEMBLY_INCOMPLETE) is added to show that the mbuf received is not reassembled p

[PATCH 4/8] security: add IPsec option for IP reassembly

A new option is added in IPsec to enable and attempt reassembly of inbound packets. Signed-off-by: Akhil Goyal --- lib/security/rte_security.h | 12 +++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/lib/security/rte_security.h b/lib/security/rte_security.h index 1228b6c8

[PATCH 5/8] app/test: add unit cases for inline IPsec offload

A new test suite is added in test app to test inline IPsec protocol offload. In this patch, a couple of predefined plain and cipher test vectors are used to verify the IPsec functionality without the need of external traffic generators. The sent packet is loopbacked onto the same interface which is

[PATCH 6/8] app/test: add IP reassembly case with no frags

test_inline_ipsec testsuite is extended to test IP reassembly of inbound fragmented packets. The fragmented packet is sent on an interface which encrypts the packet and then it is loopbacked on the same interface which decrypts the packet and then attempts IP reassembly of the decrypted packets. In

[PATCH 7/8] app/test: add IP reassembly cases with multiple fragments

More cases are added in test_inline_ipsec test suite to verify packets having multiple IP(v4/v6) fragments. These fragments are encrypted and then decrypted as per inline IPsec processing and then an attempt is made to reassemble the fragments. The reassembled packet content is matched with the kno

[PATCH 8/8] app/test: add IP reassembly negative cases

test_inline_ipsec testsuite is added with cases where the IP reassembly is incomplete and software will need to reassemble them later. The failure cases added are: - all fragments are not received. - same fragment is received more than once. - out of order fragments. Signed-off-by: Akhil Goyal --

[PATCH 0/5] net/cnxk: support IP reassembly offload

cn10k platform can support IP reassembly offload for upto 4 fragments in the inline IPsec path. The feature is enabled and tested as per [1]. The patchset depends on [1] and [2] patchsets. [1]: http://patches.dpdk.org/project/dpdk/list/?series=21052 [2]: http://patches.dpdk.org/project/dpdk/list/

[PATCH 1/5] common/cnxk: configure reassembly specific params

From: Vidya Sagar Velumuri When reassembly is enabled by application, set corresponding flags in SA during creation. Provide roc API to configure reassembly unit with active and zombie limits and step size Signed-off-by: Vidya Sagar Velumuri --- drivers/common/cnxk/cnxk_security.c | 5 -

[PATCH 3/5] net/cnxk: support IP reassembly mbuf dynfield

From: Vidya Sagar Velumuri Register the dynamic field for IPsec reassembly. Attach the fragments using the dynamic field in case of incomplete reassembly Signed-off-by: Vidya Sagar Velumuri --- drivers/net/cnxk/cn10k_ethdev_sec.c | 3 +++ drivers/net/cnxk/cn10k_rx.h | 38 +

[PATCH 4/5] net/cnxk: add dev args for min-max spi

From: Nithin Dabilpuram Dev args for setting minimum and maximum SPI value that the hardware can support and create database for SA lookup in inline IPsec processing is added. Signed-off-by: Nithin Dabilpuram --- drivers/common/cnxk/roc_nix.h | 1 + drivers/common/cnxk/roc_nix_inl.c

[PATCH 5/5] net/cnxk: add option to override outbound inline sa iv

From: Nithin Dabilpuram Add option to override outbound inline sa iv for debug purposes via environment variable. User can set env variable as: export CN10K_ETH_SEC_IV_OVR="0x0, 0x0,..." Signed-off-by: Nithin Dabilpuram --- drivers/net/cnxk/cn10k_ethdev_sec.c | 35 +

[PATCH] buildtools: fix avx512 check for Python 3.5

Python 3.5 subprocess.run() does not have a capture_output parameter (it is present only in 3.7 and up). Capture output by using subprocess.PIPE for stdout instead. Fixes: bb9cd91095b3 ("buildtools: make AVX512 check portable") Cc: sta...@dpdk.org Signed-off-by: Lance Richardson --- buildtools/b

[PATCH v1 0/3] GPU memory aligned

From: Elena Agostini Applications may need to allocate GPU memory buffers with memory address aligned to some value (e.g. page size). Similarly to the rte_malloc function, aligned size can be provided as input to rte_gpu_mem_alloc. This set of patches implements this functionality in the gpudev

[PATCH v1 1/3] gpudev: mem alloc aligned memory

From: Elena Agostini Similarly to rte_malloc, rte_gpu_mem_alloc accept as input the memory alignment size. GPU driver should return GPU memory address aligned with the input value. Signed-off-by: Elena Agostini --- lib/gpudev/gpudev.c| 10 -- lib/gpudev/gpudev_driver.h | 2 +-

[PATCH v1 2/3] app/test-gpudev: test aligned memory allocation

From: Elena Agostini Update gpudev app to test GPU memory aligned allocation. Signed-off-by: Elena Agostini --- app/test-gpudev/main.c | 13 ++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/app/test-gpudev/main.c b/app/test-gpudev/main.c index 5c1aa3d52f..f36f46cbca

[PATCH v1 3/3] gpu/cuda: mem alloc aligned memory

From: Elena Agostini Implement aligned GPU memory allocation in GPU CUDA driver. Signed-off-by: Elena Agostini --- drivers/gpu/cuda/cuda.c | 21 - 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/cuda/cuda.c b/drivers/gpu/cuda/cuda.c index 882df08e

Re: [PATCH] buildtools: fix avx512 check for Python 3.5

2022-01-03 12:09 (UTC-0500), Lance Richardson: > Python 3.5 subprocess.run() does not have a capture_output > parameter (it is present only in 3.7 and up). Capture output > by using subprocess.PIPE for stdout instead. > > Fixes: bb9cd91095b3 ("buildtools: make AVX512 check portable") > Cc: sta...@

Re: [PATCH v1 3/3] gpu/cuda: mem alloc aligned memory

On Tue, 4 Jan 2022 01:47:21 + wrote: > static int > -cuda_mem_alloc(struct rte_gpu *dev, size_t size, void **ptr) > +cuda_mem_alloc(struct rte_gpu *dev, size_t size, void **ptr, unsigned int > align) > { > CUresult res; > const char *err_string; > @@ -610,8 +612,10 @@ cuda_mem_

Re: [PATCH v1 3/3] gpu/cuda: mem alloc aligned memory

> From: Stephen Hemminger > Date: Monday, 3 January 2022 at 19:05 > To: Elena Agostini > Cc: dev@dpdk.org > Subject: Re: [PATCH v1 3/3] gpu/cuda: mem alloc aligned memory > External email: Use caution opening links or attachments> > > On Tue, 4 Jan 2022 01:47:21 + > wrote:> > > static in

Re: [PATCH v1 3/3] gpu/cuda: mem alloc aligned memory

On Mon, 3 Jan 2022 18:15:11 + Elena Agostini wrote: > > Alignment only makes sense if power of two. The code should check that and > > optimize > > for that. > > > > The alignment value is checked in the gpudev library before > passing it to the driver. > > Adding this kind of checks in

[Bug 919] Error attaching device to DPDK

https://bugs.dpdk.org/show_bug.cgi?id=919 Bug ID: 919 Summary: Error attaching device to DPDK Product: DPDK Version: 19.11 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: N

Re: [PATCH v1 3/3] gpu/cuda: mem alloc aligned memory

> On Mon, 3 Jan 2022 18:15:11 + > Elena Agostini wrote: > > > > Alignment only makes sense if power of two. The code should check that > > > and optimize > > > for that. > > > > > > > The alignment value is checked in the gpudev library before > > passing it to the driver. > > > > Adding this

[PATCH v1] gpudev: pin GPU memory

From: Elena Agostini Enable the possibility to make a GPU memory area accessible from the CPU. GPU memory has to be allocated via rte_gpu_mem_alloc(). This patch allows the gpudev library to pin, through the GPU driver, a chunk of GPU memory and to return a memory pointer usable by the CPU to a

[PATCH v2] gpudev: pin GPU memory

From: Elena Agostini Enable the possibility to make a GPU memory area accessible from the CPU. GPU memory has to be allocated via rte_gpu_mem_alloc(). This patch allows the gpudev library to pin, through the GPU driver, a chunk of GPU memory and to return a memory pointer usable by the CPU to a

[PATCH v2] gpudev: pin GPU memory

From: Elena Agostini Enable the possibility to make a GPU memory area accessible from the CPU. GPU memory has to be allocated via rte_gpu_mem_alloc(). This patch allows the gpudev library to pin, through the GPU driver, a chunk of GPU memory and to return a memory pointer usable by the CPU to a

[PATCH] net/mlx5: fix risk in Rx descriptor read in NEON vector path

In NEON vector PMD, vector load loads two contiguous 8B of descriptor data into vector register. Given vector load ensures no 16B atomicity, read of the word that includes op_own field could be reordered after read of other words. In this case, some words could contain invalid data. Reloaded qword