[dpdk-dev] [PATCH] vhost: add back support for concurrent enqueue

2016-08-16 Thread Yuanhan Liu
On Mon, Aug 15, 2016 at 01:00:24PM -0700, Rich Lane wrote: > Concurrent enqueue is an important performance optimization when the number > of cores used for switching is different than the number of vhost queues. > I've observed a 20% performance improvement compared to a strategy that > binds

[dpdk-dev] vhost [query] : support for multiple ports and non VMDQ devices in vhost switch

2016-08-16 Thread Yuanhan Liu
On Tue, Aug 09, 2016 at 04:42:33PM +0530, Pankaj Chauhan wrote: > > Hi, > > I am working on an NXP platform where we intend to use user space vhost > switch (examples/vhost) as backend for VIRTIO devices. But there are two > limitations in current vhost-switch (examples/vhost)that are

[dpdk-dev] [dpdk-users] rte_zmalloc() returning non-zeroed memory on FreeBSD

2016-08-16 Thread Sergio Gonzalez Monroy
On 15/08/2016 18:23, Harris, James R wrote: > >> -Original Message- >> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Thomas Monjalon >> Sent: Thursday, August 11, 2016 12:05 AM >> To: users at dpdk.org; dev at dpdk.org; Gonzalez Monroy, Sergio; Richardson, >> Bruce >> Cc:

[dpdk-dev] [PATCH v5 4/8] table: enable table library for ppc64le

2016-08-16 Thread Chao Zhu
Gowrishankar, The unit test of table got failed. When I took a look at the code, it failed on this line in file lib/librte_table/rte_table_hash_key8.c , lib/librte_table/rte_table_hash_key16.c, lib/librte_table/rte_table_hash_key32.c: if ((check_params_create_lru(p) != 0) ||

[dpdk-dev] [PATCH] kni: error rollback with kni_dev_remove could cause a kernel crash

2016-08-16 Thread zhouyangchao

[dpdk-dev] [PATCH] net/i40e: fix dropping packets with Ethertype 0x88A8

2016-08-16 Thread Beilei Xing
Refer to FW default setting, Ethertype 0x88A8 is treated as S-TAG, packects with S-TAG should be received in Port Virtualizer mode, but Port Virtual mode is not initialized in DPDK. So X710 will packets with Ethertype 0x88A8. This patch fixes this issue by turning off S-TAG identification. Fixes:

[dpdk-dev] [PATCH v6 0/9] enable lpm, acl and other missing libraries in ppc64le

2016-08-16 Thread Gowrishankar Muthukrishnan
This patchset enables LPM, ACL and other few missing libs in ppc64le and also address few patches in related examples (ip_pipeline and l3fwd). Test report: 1. LPM and ACL unit tests passed. Steps: compile test app and run (with any needed params) lpm_autotest acl_autotest

[dpdk-dev] [PATCH v6 2/9] acl: add altivec intrinsics for dpdk acl on ppc_64

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch adds port for ACL library in ppc64le. Signed-off-by: Gowrishankar Muthukrishnan --- app/test-acl/main.c | 4 + config/defconfig_ppc_64-power8-linuxapp-gcc | 1 - lib/librte_acl/Makefile | 2 + lib/librte_acl/acl.h

[dpdk-dev] [PATCH v6 3/9] l3fwd: add altivec support for em_hash_key

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch adds ppc64le port for em_mask_key function. Signed-off-by: Gowrishankar Muthukrishnan --- examples/l3fwd/l3fwd_em.c | 10 +- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/examples/l3fwd/l3fwd_em.c b/examples/l3fwd/l3fwd_em.c index def5a02..6053a62 100644 ---

[dpdk-dev] [PATCH v6 5/9] sched: enable sched library for ppc64le

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch enables librte_sched in ppc64le. Signed-off-by: Gowrishankar Muthukrishnan --- config/defconfig_ppc_64-power8-linuxapp-gcc | 1 - 1 file changed, 1 deletion(-) diff --git a/config/defconfig_ppc_64-power8-linuxapp-gcc b/config/defconfig_ppc_64-power8-linuxapp-gcc index

[dpdk-dev] [PATCH v6 6/9] port: enable port library for ppc64le

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch enables librte_port in ppc64le. Signed-off-by: Gowrishankar Muthukrishnan --- config/defconfig_ppc_64-power8-linuxapp-gcc | 1 - 1 file changed, 1 deletion(-) diff --git a/config/defconfig_ppc_64-power8-linuxapp-gcc b/config/defconfig_ppc_64-power8-linuxapp-gcc index

[dpdk-dev] [PATCH v6 7/9] pipeline: enable pipeline library for ppc64le

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch enables librte_pipeline for ppc64le. Signed-off-by: Gowrishankar Muthukrishnan --- config/defconfig_ppc_64-power8-linuxapp-gcc | 2 -- 1 file changed, 2 deletions(-) diff --git a/config/defconfig_ppc_64-power8-linuxapp-gcc b/config/defconfig_ppc_64-power8-linuxapp-gcc index

[dpdk-dev] [PATCH v6 4/9] table: enable table library for ppc64le

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch enables librte_table in ppc64le. Signed-off-by: Gowrishankar Muthukrishnan --- config/defconfig_ppc_64-power8-linuxapp-gcc | 1 - 1 file changed, 1 deletion(-) diff --git a/config/defconfig_ppc_64-power8-linuxapp-gcc b/config/defconfig_ppc_64-power8-linuxapp-gcc index

[dpdk-dev] [PATCH v6 8/9] ip_pipeline: fix lcore mapping for varying SMT threads as in ppc64

2016-08-16 Thread Gowrishankar Muthukrishnan
This patch fixes ip_pipeline panic in app_init_core_map while preparing cpu core map in powerpc with SMT off. cpu_core_map_compute_linux currently prepares core mapping based on file existence in sysfs ie. /sys/devices/system/cpu/cpu/topology/physical_package_id

[dpdk-dev] [PATCH v6 9/9] table: align rte table hash structs for cache line size

2016-08-16 Thread Gowrishankar Muthukrishnan
rte table hash structs rte_bucket_4_8, rte_bucket_4_16 and rte_bucket_4_32 have to be cache aligned as required by their corresponding hash create functions rte_table_hash_create_key8_lru etc. Signed-off-by: Gowrishankar Muthukrishnan --- lib/librte_table/rte_table_hash_key16.c | 4 ++--

[dpdk-dev] [PATCH] optimize vhost enqueue

2016-08-16 Thread Zhihong Wang
This patch optimizes the vhost enqueue function: rte_vhost_enqueue_burst. Currently there're 2 callbacks for vhost enqueue: * virtio_dev_merge_rx for mrg_rxbuf turned on cases. * virtio_dev_rx for mrg_rxbuf turned off cases. The virtio_dev_merge_rx doesn't provide optimal performance, also

[dpdk-dev] [PATCH] contigmem: zero all pages during mmap

2016-08-16 Thread Sergio Gonzalez Monroy
On 15/08/2016 19:17, Jim Harris wrote: > On Linux, all huge pages are zeroed by the kernel before > first access by the DPDK application. But on FreeBSD, > the contigmem driver would only zero the contiguous > memory regions during initial driver load. > > DPDK commit b78c91751 eliminated the

[dpdk-dev] [PATCH] optimize vhost enqueue

2016-08-16 Thread Maxime Coquelin
Hi Zhihong, On 08/16/2016 05:50 AM, Zhihong Wang wrote: > This patch optimizes the vhost enqueue function: rte_vhost_enqueue_burst. > > Currently there're 2 callbacks for vhost enqueue: > * virtio_dev_merge_rx for mrg_rxbuf turned on cases. > * virtio_dev_rx for mrg_rxbuf turned off cases. >

[dpdk-dev] [PATCH] Performance optimization of ACL build process

2016-08-16 Thread Vladyslav Buslov
Hello, In our application we need to be able to allocate tens of thousands of ACLs at runtime. Testing revealed significant performance problems. We were able to track them to memset in calloc function which caused multiple page walks per invocation. Modifying tb_mem to use huge page memory

[dpdk-dev] [PATCH] acl: use rte_calloc for temporary memory allocation

2016-08-16 Thread Vladyslav Buslov
Acl build process uses significant amount of memory which degrades performance by causing page walks when memory is allocated on regular heap using libc calloc. This commit changes tb_mem to allocate temporary memory on huge pages with rte_calloc. Signed-off-by: Vladyslav Buslov ---

[dpdk-dev] [PATCH] nfp: using random mac address if not a configured mac

2016-08-16 Thread Alejandro Lucero
Signed-off-by: Alejandro Lucero --- drivers/net/nfp/nfp_net.c | 28 1 file changed, 16 insertions(+), 12 deletions(-) diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c index 82e3e4e..45d122d 100644 --- a/drivers/net/nfp/nfp_net.c +++

[dpdk-dev] [PATCH] nfp: fixing bug when copying mac address

2016-08-16 Thread Alejandro Lucero
Fixes: defb9a5dd156 ("nfp: introduce driver initialization") Signed-off-by: Alejandro Lucero --- drivers/net/nfp/nfp_net.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c index 45d122d..229c8e6 100644 ---

[dpdk-dev] [PATCH] nfp: unregister interrupt callback function when closing device

2016-08-16 Thread Alejandro Lucero
With an app using hotplug feature, when a device is unplugged without unregistering makes the interrupt handling unstable. Fixes: 6c53f87b3497 ("nfp: add link status interrupt") Signed-off-by: Alejandro Lucero --- drivers/net/nfp/nfp_net.c | 5 + 1 file changed, 5 insertions(+) diff --git

[dpdk-dev] [PATCH] kni: add module parameter 'bind_to_core'

2016-08-16 Thread Vladyslav Buslov
Allow binding KNI thread to specific core in single threaded mode. Signed-off-by: Vladyslav Buslov --- lib/librte_eal/linuxapp/kni/kni_misc.c | 13 - 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/lib/librte_eal/linuxapp/kni/kni_misc.c

[dpdk-dev] [PATCH v2] contigmem: zero all pages during mmap

2016-08-16 Thread Jim Harris
On Linux, all huge pages are zeroed by the kernel before first access by the DPDK application. But on FreeBSD, the contigmem driver would only zero the contiguous memory regions during initial driver load. DPDK commit b78c91751 eliminated the explicit memset() operation for rte_zmalloc(), which

[dpdk-dev] [dpdk-users] rte_zmalloc() returning non-zeroed memory on FreeBSD

2016-08-16 Thread Harris, James R
> -Original Message- > From: Gonzalez Monroy, Sergio > Sent: Tuesday, August 16, 2016 12:37 AM > To: Harris, James R; Thomas Monjalon; users at dpdk.org; dev at dpdk.org; > Richardson, Bruce > Cc: Verkamp, Daniel > Subject: Re: [dpdk-dev] [dpdk-users] rte_zmalloc() returning non-zeroed >

[dpdk-dev] [PATCH 0/2] examples/vhost: rename a CLI option and support multiple socket files

2016-08-16 Thread Jiayu Hu
Patch 1: rename the CLI option "dev-basename". Patch 2: add multiple socket files support. Jiayu Hu (2): examples/vhost: rename dev-basename examples/vhost: support multiple socket files examples/vhost/main.c | 81 ++- 1 file changed, 54

[dpdk-dev] [PATCH 1/2] examples/vhost: rename dev-basename

2016-08-16 Thread Jiayu Hu
In examples/vhost, "dev-basename" is a program option, which is to set the vhost-net socket used by vhost-user, or the character device used by vhost-cuse. Since vhost-cuse should be dropped, and "dev-basename" is not a suitable name for the vhost-net socket. Therefore, this patch is to change