[PATCH v1 2/2] net/axgbe: move offloads to Rx/Tx queue setup

For Multiprocess dpdk applications retrieving the offload parameter for receive packet directly from struct eth_dev giving segmentation fault since rxmode.offloads from eth_dev was null duirng recv_pkts. So retrieved offload from rx/tx_queue_setup() Signed-off-by: Jesna K E --- drivers/net/axgbe

[PATCH v1 1/2] net/axgbe: add multi-process support

Added multi-process support for axgbe PMD To achieve multi-process support separate out TX and RX function inside the axgbe driver and call that from a secondary process when it is attaching to an already-configured NIC Signed-off-by: Jesna K E --- .mailmap | 1 +

[PATCH v3] bus/fslmc: unlock spinlock in case of timeout

Bug in function mc_send_command, A spinlock is not unlocked in case of timeouts. Fixes: 79711846f676 ("bus/fslmc: add timeout in MC send command") Cc: sta...@dpdk.org Signed-off-by: Alvaro Karsz Reviewed-by: David Marchand -- v2: - Fix code indentation. - Fix "Fixes: " tag forma

回复: 回复: 回复: 回复: [PATCH v2 1/3] ethdev: add API for direct rearm mode

Hi, Konstantin > -邮件原件- > 发件人: Konstantin Ananyev > 发送时间: Tuesday, November 1, 2022 12:37 AM > 收件人: Feifei Wang ; tho...@monjalon.net; > Ferruh Yigit ; Andrew Rybchenko > ; Ray Kinsella > 抄送: dev@dpdk.org; nd ; Honnappa Nagarahalli > ; Ruifeng Wang > > 主题: Re: 回复: 回复: 回复: [PATCH v2 1/3]

[PATCH v3 3/3] net/ixgbe: enable direct rearm with separate API

Add internal API to separate direct rearm operations between Rx and Tx. Signed-off-by: Feifei Wang Reviewed-by: Ruifeng Wang Reviewed-by: Honnappa Nagarahalli --- drivers/net/ixgbe/ixgbe_ethdev.c | 1 + drivers/net/ixgbe/ixgbe_ethdev.h | 3 ++ drivers/net/ixgbe/ixgbe_rxtx.c

[PATCH v3 2/3] net/i40e: enable direct rearm with separate API

Add internal API to separate direct rearm operations between Rx and Tx. Suggested-by: Honnappa Nagarahalli Signed-off-by: Feifei Wang Reviewed-by: Ruifeng Wang Reviewed-by: Honnappa Nagarahalli --- drivers/net/i40e/i40e_ethdev.c | 1 + drivers/net/i40e/i40e_ethdev.h | 2 +

[PATCH v3 1/3] ethdev: enable direct rearm with separate API

Add 'tx_fill_sw_ring' and 'rx_flush_descriptor' API into direct rearm mode for separate Rx and Tx Operation. And this can support different multiple sources in direct rearm mode. For examples, Rx driver is ixgbe, and Tx driver is i40e. Suggested-by: Honnappa Nagarahalli Suggested-by: Ruifeng Wang

[PATCH v3 0/3] Direct re-arming of buffers on receive side

Currently, the transmit side frees the buffers into the lcore cache and the receive side allocates buffers from the lcore cache. The transmit side typically frees 32 buffers resulting in 32*8=256B of stores to lcore cache. The receive side allocates 32 buffers and stores them in the receive side so

[PATCH] cryptodev: increase default max nb queue pairs

Increase default maximum number of queue pairs to 64. For sw crypto PMDs, the number of queue pairs can be increased as required. The default value (which is set to 8), prevents usage of sw cryptodev PMDs with larger number of cores. Application would be allowed to specify the number of queue pair

[PATCH v6 4/4] eventdev/timer: change eventdev reconfig logic

When rte_event_timer_adapter_create() is used for creating adapter instance, eventdev is reconfigured with additional ``rte_event_dev_config::nb_event_ports`` parameter. This eventdev reconfig logic is enhanced to increment the ``rte_event_dev_config::nb_single_link_event_port_queues`` parameter i

[PATCH v6 3/4] eventdev/crypto: change eventdev reconfig logic

When rte_event_crypto_adapter_create() is used for creating adapter instance, eventdev is reconfigured with additional ``rte_event_dev_config::nb_event_ports`` parameter. This eventdev reconfig logic is enhanced to increment the ``rte_event_dev_config::nb_single_link_event_port_queues`` parameter

[PATCH v6 2/4] eventdev/eth_tx: change eventdev reconfig logic

When rte_event_eth_tx_adapter_create() is used for creating adapter instance, eventdev is reconfigured with additional ``rte_event_dev_config::nb_event_ports`` parameter. This eventdev reconfig logic is enhanced to increment the ``rte_event_dev_config::nb_single_link_event_port_queues`` parameter

[PATCH v6 1/4] eventdev/eth_rx: change eventdev reconfig logic

When rte_event_eth_rx_adapter_create() or rte_event_eth_rx_adapter_create_with_params() is used for creating adapter instance, eventdev is reconfigured with additional ``rte_event_dev_config::nb_event_ports`` parameter. This eventdev reconfig logic is enhanced to increment the ``rte_event_dev_conf

[PATCH v2] cryptodev: add algo enums to string conversion APIs

Symmetric/Asymmetric algorithm strings are accessed by application using arrays in cryptodev lib, which hampers new algorithms addition in the array due to ABI breakage. These arrays are now deprecated and will be removed in next ABI break release. New APIs are added for getting the algorithm strin

[PATCH] common/cnxk: reduce channel count per LMAC

From: Sunil Kumar Kori Due to limitation, imposed by latest kernel, maximum number of channel supported per LMAC is reduced to 8. Because of this change application gets failed to initialize for more channels. Also this limitation impacts PFC functional behaviour. So patch just aligns the suppo

[PATCH v2] vfio: do not coalesce DMA mappings

At the cleanup time when dma unmap is done, linux kernel does not allow unmap of individual segments which were coalesced together while creating the DMA map for type1 IOMMU mappings. So, this change updates the mapping of the memory segments(hugepages) on a per-page basis. Signed-off-by: Nipun Gu

[PATCH v2 6/6] app/dumpcap: support interface name and description

Support setting --ifname and --ifdescr options to record information in the start of the pcapng interface description block. Also, records filter (if any) used in the file. This also makes sure only the interfaces being recorded in the capture file are in the interface block. Signed-off-by: Steph

[PATCH v2 4/6] app/dumpcap: support multiple interfaces

The code to handle multiple interfaces did not work before. Need to initialize the list of interfaces while parsing the command line. Each interface should the ability to have its own options. Fixes another bug where dumpcap would always disable promiscious on exit, even if set by the application.

[PATCH v2 5/6] pcapng: require per-interface information

This changes the API for how pcapng is used. Before each interface was automatically added to the capture file. Now the application must add each interface. Note: API changes are allowed because this is an experimental interface. This allows application to specify extra meta data like interface n

[PATCH v2 3/6] app/dumpcap: check for invalid interface name

Avoid any possible issue with ridiculously long name. Signed-off-by: Stephen Hemminger --- app/dumpcap/main.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/app/dumpcap/main.c b/app/dumpcap/main.c index 1c47ff851252..a7aec021204a 100644 --- a/app/dumpcap/main.c +++ b/app/dumpcap/main.c @

[PATCH v2 1/6] app/dumpcap: fix storing port identifier

From: Ben Magistro When dumpcap adds an interface, the port was not being preserved. This results in the structure being initialized and the port field being set to 0 regardless of what port was actually selected. This unset field is then used in both the enable and cleanup calls. This could resu

[PATCH v2 2/6] app/dumpcap: remove unused variable

The port2intf variable was set but never used. Signed-off-by: Stephen Hemminger --- app/dumpcap/main.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/app/dumpcap/main.c b/app/dumpcap/main.c index 4751ca26b892..1c47ff851252 100644 --- a/app/dumpcap/main.c +++ b/app/dumpcap/main.c @@ -90,7 +

[PATCH v2 0/6] dumpcap support multiple interfaces

This set of patches fixed the handling of multiple interfaces in dumpcap. It also supports recording the name, description and capture filter in the pcapng file. The last two are not necessary for stable. Ben Magistro (1): app/dumpcap: fix storing port identifier Stephen Hemminger (5): app/d

[PATCH] vhost: fix net header settings in vhost datapath

In vhost sync batch enqueue, the "num_buffers" of virtio net header is not be set, but it should be set to 1 if the mrg_rxbuf feature is turned on, This patch fix the issue. Fixes: ef861692c398 ("vhost: add packed ring batch enqueue") Cc: sta...@dpdk.org Signed-off-by: Wenwu Ma --- lib/vhost/vi

RE: [PATCH] app/testpmd: fix secondary process not forwarding

>> Under multi-process scenario, the secondary process gets queue state >> from the wrong location (the global variable 'ports'). Therefore, the >> secondary process can not forward since "stream_init" is not called. >> >> This commit fixes the issue by calling 'rte_eth_rx/tx_queue_info_get' >> to

[PATCH v2 1/1] eal/linux: reject --huge-dir not parent of mountpt

The code added for allowing --huge-dir to specify hugetlbfs sub-directories has a bug where it incorrectly matches mounts that contain a prefix of the specified --huge-dir. Consider --huge-dir=/dev/hugepages1G is passed to rte_eal_init. Given the following hugetlbfs mounts $ mount | grep hugetlbf

Re: [PATCH 1/1] eal/linux: reject mountpt shorter than --huge-dir

Please ignore this patch, I'll submit an updated one. I somehow managed to only execute a subset of the fast-tests suite initially and didn't run eal_flags_misc_autotest at all. Now I see that my proposed fix is flawed, I will submit another try soon. Sorry for the noise On Tue, Jan 3, 2023 at 1

RE: [PATCH v1 0/3] bbdev: remove offload cost

Hi Maxime, Happy new year. What does the state " Awaiting Upstream" mean in patchwork for that serie? Can this be applied? Thanks Nic > -Original Message- > From: Chautru, Nicolas > Sent: Wednesday, December 14, 2022 3:34 PM > To: dev@dpdk.org; maxime.coque...@redhat.com > Cc: Var

Re: [PATCH v2] bus/fslmc: unlock spinlock in case of timeout

Hi David, Thanks for your review. I'll create a new version without the comment, and will cc sta...@dpdk.org.

[PATCH 1/1] eal/linux: reject mountpt shorter than --huge-dir

The code added for allowing --huge-dir to specify hugetlbfs sub-directories has a bug where it incorrectly matches mounts that contain a prefix of the specified --huge-dir. Consider --huge-dir=/dev/hugepages1G is passed to rte_eal_init. Given the following hugetlbfs mounts $ mount | grep hugetlbf

Re: [PATCH v2] net/i40e: don't check link status on device start

Hi i40e maintainers, On Tue, Dec 13, 2022 at 10:19 AM David Marchand wrote: > > The mentioned changes broke existing applications when the link status > of i40e ports is down at the time the port is started. > Revert those changes, the original issue will need a different fix. > > Fixes: a4ba7736

[PATCH v3] malloc: enhance NUMA affinity heuristic

Trying to allocate memory on the first detected numa node has less chance to find some memory actually available rather than on the main lcore numa node (especially when the DPDK application is started only on one numa node). Signed-off-by: David Marchand --- Changes since v2: - add uncommitted l

[PATCH] eventdev/eth_rx: fix return of rx adapter instance get

The API to get rx adapter instance is returning error for event devices with internal port capability and eth_rx_adapter_instance_get op undefined. But as the Rx adapter is internally maintaining the queue information needed to find the instance id, event devices can opt out from defining the op. M

Re: [PATCH v3] kni: fix possible alloc_q starvation when mbufs are exhausted

On 12/30/2022 4:23 AM, Yangchao Zhou wrote: > In some scenarios, mbufs returned by rte_kni_rx_burst are not freed > immediately. So kni_allocate_mbufs may be failed, but we don't know. > > Even worse, when alloc_q is completely exhausted, kni_net_tx in > rte_kni.ko will drop all tx packets. kni_al

Re: Strange behavior with rte_pktmbuf_clone call

On 12/23/2022 6:51 AM, NAGENDRA BALAGANI wrote: > Hi, > >   > > I am seeing strange behavior where rte_pktmbuf_clone is not giving > desired result. > Hi Nagendra, What is the desired result? 'rte_pktmbuf_clone()' creates indirect mbuf [1], which means network packet is shared between mbufs,

[PATCH] net/mlx5: fix read device clock in real time mode

Since ConnectX-6DX the real time timestamp mode is supported. The rte_eth_read_clock() routine queries current timestamp value from the PMD. The mlx5 PMD has special infrastructure to schedule packet sending in real time mode which can be engaged with tx_pp devarg. This infrastructure provides the

[PATCH] net/mlx5: fix irrelevant send skew devarg warning

Since ConnectX-7 hardware supports the send scheduling on explicitly specified timestamp, if this feature support is detected by PMD the tx_skew devarg becomes applicable and appropriate warning should not be emitted. Fixes: 49e8797619e40 ("net/mlx5: support wait on time in Tx") Cc: sta...@dpdk.or

[PATCH] net/mlx5: fix wait descriptor opcode for ConnectX-7

Since ConnectX-7 the special hardware capability to schedule packet sending on specified time is provided. It uses special WAIT WQE (hardware descriptor) with wait condition defined by opcode. The CYCLIC_BIGGER was wrongly used instead of CYCLIC_SMALLER. This caused arbitrary and stale time shift i

[PATCH v2] malloc: enhance NUMA affinity heuristic

Trying to allocate memory on the first detected numa node has less chance to find some memory actually available rather than on the main lcore numa node (especially when the DPDK application is started only on one numa node). Signed-off-by: David Marchand --- Changes since v1: - accomodate for co

Re: [PATCH] malloc: enhance NUMA affinity heuristic

On Tue, Dec 27, 2022 at 10:00 AM zhoumin wrote: > > Hi David, > > > First of all, I sincerely apologize for the late reply. > > I had checked this issue carefully and had some useful findings. > > On Wed, Dec 21, 2022 at 22:57 PM, David Marchand wrote: > > Hello Min, > > > > On Wed, Dec 21, 2022 a

[PATCH] gpudev: fix deadlocks when registering callback

gpu_callback_lock was not released in some branches of the register helper. While at it, set rte_errno in one branch. Fixes: 18cb07563165 ("gpudev: add event notification") Cc: sta...@dpdk.org Signed-off-by: David Marchand --- lib/gpudev/gpudev.c | 5 - 1 file changed, 4 insertions(+), 1 de

[PATCH] fbarray: fix metadata dump

If the passed fbarray is invalid, its lock was not taken before releasing. Fixes: c44d09811b40 ("eal: add shared indexed file-backed array") Cc: sta...@dpdk.org Signed-off-by: David Marchand --- lib/eal/common/eal_common_fbarray.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --

Re: [PATCH v2] bus/fslmc: unlock spinlock in case of timeout

On Mon, Dec 26, 2022 at 8:47 AM Alvaro Karsz wrote: > > Bug in function mc_send_command, > A spinlock is not unlocked in case of timeouts. > > Fixes: 79711846f676 ("bus/fslmc: add timeout in MC send command") This is worth backporting to 22.11 LTS. Please add: Cc: sta...@dpdk.org > Signed-off-by

RE: [PATCH v2 1/3] ethdev: add ICMPv6 ID and sequence

Hi Leo, > -Original Message- > From: Leo Xu (Networking SW) > Sent: Tuesday, 20 December 2022 9:44 > > This patch adds API support for ICMPv6 ID and sequence. > 1: Add two new pattern item types for ICMPv6 echo request and reply: > RTE_FLOW_ITEM_TYPE_ICMP6_ECHO_REQUEST > RTE_FLOW_IT