* [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux
@ 2016-01-22 16:00 Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 1/2] kdp: add kernel data path kernel module Ferruh Yigit
` (2 more replies)
0 siblings, 3 replies; 7+ messages in thread
From: Ferruh Yigit @ 2016-01-22 16:00 UTC (permalink / raw)
To: dev
This is slow data path communication implementation based on existing KNI.
Difference is: librte_kni converted into a PMD, kdp kernel module is almost
same except all control path functionality removed and some simplification done.
Motivation is to simplify slow path data communication.
Now any application can use this new PMD to send/get data to Linux kernel.
PMD initialization functions handles creating virtual interfaces (with help of
kdp kernel module) and created FIFO. FIFO is used to share data between
userspace and kernelspace.
Sample usage:
1) Transfer any packet received from NIC that bound to DPDK, to the Linux kernel
a) insert kdp kernel module
insmod build/kmod/rte_kdp.ko
b) bind NIC to the DPDK using dpdk_nic_bind.py
c) ./testpmd --vdev eth_kdp0
c1) testpmd show two ports, one of them physical, other virtual
...
Configuring Port 0 (socket 0)
Port 0: 00:00:00:00:00:00
Configuring Port 1 (socket 0)
...
Checking link statuses...
Port 0 Link Up - speed 10000 Mbps - full-duplex
Port 1 Link Up - speed 10000 Mbps - full-duplex
Done
c2) This will create "kdp0" Linux interface
$ ip l show kdp0
21: kdp0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
d) Linux port can be used for data
d1)
$ ifconfig kdp0 1.0.0.2
$ ping 1.0.0.1
PING 1.0.0.1 (1.0.0.1) 56(84) bytes of data.
64 bytes from 1.0.0.1: icmp_seq=1 ttl=64 time=0.789 ms
64 bytes from 1.0.0.1: icmp_seq=2 ttl=64 time=0.881 ms
d2)
$ tcpdump -nn -i kdp0
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on kdp0, link-type EN10MB (Ethernet), capture size 262144 bytes
15:01:22.407506 IP 1.0.0.1 > 1.0.0.2: ICMP echo request, id 40016, seq 18, length 64
15:01:22.408521 IP 1.0.0.2 > 1.0.0.1: ICMP echo reply, id 40016, seq 18, length 64
2) Data travels between virtual Linux interfaces pass from DPDK application,
application can alter data
a) insert kdp kernel module
insmod build/kmod/rte_kdp.ko
b) No physical NIC involved
c) ./testpmd --vdev eth_kdp0 --vdev eth_kdp1
c1) testpmd show two ports, both of them are virtual
...
Configuring Port 0 (socket 0)
Port 0: 00:00:00:00:00:00
Configuring Port 1 (socket 0)
Port 1: 00:00:00:00:00:00
Checking link statuses...
Port 0 Link Up - speed 10000 Mbps - full-duplex
Port 1 Link Up - speed 10000 Mbps - full-duplex
Done
c2) This will create "kdp0" and "kdp1" Linux interfaces
$ ip l show kdp0; ip l show kdp1
22: kdp0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
23: kdp1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
d) Data travel between virtual ports pass from DPDK application
$ifconfig kdp0 1.0.0.1
$ifconfig kdp1 1.0.0.2
d1)
$ ping 1.0.0.1
PING 1.0.0.1 (1.0.0.1) 56(84) bytes of data.
64 bytes from 1.0.0.1: icmp_seq=1 ttl=64 time=3.57 ms
64 bytes from 1.0.0.1: icmp_seq=2 ttl=64 time=1.85 ms
64 bytes from 1.0.0.1: icmp_seq=3 ttl=64 time=1.89 ms
d2)
$ tcpdump -nn -i kdp0
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on kdp0, link-type EN10MB (Ethernet), capture size 262144 bytes
15:20:51.908543 IP 1.0.0.2 > 1.0.0.1: ICMP echo request, id 41234, seq 1, length 64
15:20:51.909570 IP 1.0.0.1 > 1.0.0.2: ICMP echo reply, id 41234, seq 1, length 64
15:20:52.909551 IP 1.0.0.2 > 1.0.0.1: ICMP echo request, id 41234, seq 2, length 64
15:20:52.910577 IP 1.0.0.1 > 1.0.0.2: ICMP echo reply, id 41234, seq 2, length 64
Ferruh Yigit (2):
kdp: add kernel data path kernel module
kdp: add virtual PMD for kernel slow data path communication
config/common_linuxapp | 9 +-
drivers/net/Makefile | 3 +-
drivers/net/kdp/Makefile | 60 +++
drivers/net/kdp/rte_eth_kdp.c | 405 +++++++++++++++
drivers/net/kdp/rte_kdp.c | 365 +++++++++++++
drivers/net/kdp/rte_kdp.h | 113 ++++
drivers/net/kdp/rte_kdp_fifo.h | 91 ++++
drivers/net/kdp/rte_pmd_kdp_version.map | 4 +
lib/librte_eal/common/include/rte_log.h | 3 +-
lib/librte_eal/linuxapp/Makefile | 5 +-
lib/librte_eal/linuxapp/eal/Makefile | 3 +-
.../linuxapp/eal/include/exec-env/rte_kdp_common.h | 138 +++++
lib/librte_eal/linuxapp/kdp/Makefile | 56 ++
lib/librte_eal/linuxapp/kdp/compat.h | 29 ++
lib/librte_eal/linuxapp/kdp/kdp_dev.h | 85 +++
lib/librte_eal/linuxapp/kdp/kdp_fifo.h | 94 ++++
lib/librte_eal/linuxapp/kdp/kdp_misc.c | 463 +++++++++++++++++
lib/librte_eal/linuxapp/kdp/kdp_net.c | 578 +++++++++++++++++++++
mk/rte.app.mk | 3 +-
19 files changed, 2501 insertions(+), 6 deletions(-)
create mode 100644 drivers/net/kdp/Makefile
create mode 100644 drivers/net/kdp/rte_eth_kdp.c
create mode 100644 drivers/net/kdp/rte_kdp.c
create mode 100644 drivers/net/kdp/rte_kdp.h
create mode 100644 drivers/net/kdp/rte_kdp_fifo.h
create mode 100644 drivers/net/kdp/rte_pmd_kdp_version.map
create mode 100644 lib/librte_eal/linuxapp/eal/include/exec-env/rte_kdp_common.h
create mode 100644 lib/librte_eal/linuxapp/kdp/Makefile
create mode 100644 lib/librte_eal/linuxapp/kdp/compat.h
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_dev.h
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_fifo.h
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_misc.c
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_net.c
--
2.5.0
^ permalink raw reply [flat|nested] 7+ messages in thread
* [dpdk-dev] [RFC 1/2] kdp: add kernel data path kernel module
2016-01-22 16:00 [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Ferruh Yigit
@ 2016-01-22 16:00 ` Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 2/2] kdp: add virtual PMD for kernel slow data path communication Ferruh Yigit
2016-01-22 16:31 ` [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Thomas Monjalon
2 siblings, 0 replies; 7+ messages in thread
From: Ferruh Yigit @ 2016-01-22 16:00 UTC (permalink / raw)
To: dev
This kernel module is based on KNI module, but this one is stripped
version of it and only for data messages, no control functionality
provided.
FIFO implementation of the KNI is kept exact same, but ethtool related
code removed and virtual network management related code simplified.
This module contains kernel support to create network devices and
this module has a simple driver for virtual network device, the driver
simply puts/gets packages to/from FIFO instead of real hardware.
FIFO is created owned by userspace application, which is for this case
KDP PMD.
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
config/common_linuxapp | 8 +-
lib/librte_eal/linuxapp/Makefile | 5 +-
lib/librte_eal/linuxapp/eal/Makefile | 3 +-
.../linuxapp/eal/include/exec-env/rte_kdp_common.h | 138 +++++
lib/librte_eal/linuxapp/kdp/Makefile | 56 ++
lib/librte_eal/linuxapp/kdp/compat.h | 29 ++
lib/librte_eal/linuxapp/kdp/kdp_dev.h | 85 +++
lib/librte_eal/linuxapp/kdp/kdp_fifo.h | 94 ++++
lib/librte_eal/linuxapp/kdp/kdp_misc.c | 463 +++++++++++++++++
lib/librte_eal/linuxapp/kdp/kdp_net.c | 578 +++++++++++++++++++++
10 files changed, 1456 insertions(+), 3 deletions(-)
create mode 100644 lib/librte_eal/linuxapp/eal/include/exec-env/rte_kdp_common.h
create mode 100644 lib/librte_eal/linuxapp/kdp/Makefile
create mode 100644 lib/librte_eal/linuxapp/kdp/compat.h
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_dev.h
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_fifo.h
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_misc.c
create mode 100644 lib/librte_eal/linuxapp/kdp/kdp_net.c
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 74bc515..73c91d8 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -1,6 +1,6 @@
# BSD LICENSE
#
-# Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
+# Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
@@ -320,6 +320,12 @@ CONFIG_RTE_LIBRTE_PMD_XENVIRT=n
CONFIG_RTE_LIBRTE_PMD_NULL=y
#
+# Compile KDP PMD
+#
+CONFIG_RTE_KDP_KMOD=y
+CONFIG_RTE_KDP_PREEMPT_DEFAULT=y
+
+#
# Do prefetch of packet data within PMD driver receive function
#
CONFIG_RTE_PMD_PACKET_PREFETCH=y
diff --git a/lib/librte_eal/linuxapp/Makefile b/lib/librte_eal/linuxapp/Makefile
index d9c5233..e3f91a7 100644
--- a/lib/librte_eal/linuxapp/Makefile
+++ b/lib/librte_eal/linuxapp/Makefile
@@ -1,6 +1,6 @@
# BSD LICENSE
#
-# Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+# Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
@@ -38,6 +38,9 @@ DIRS-$(CONFIG_RTE_LIBRTE_EAL_LINUXAPP) += eal
ifeq ($(CONFIG_RTE_KNI_KMOD),y)
DIRS-$(CONFIG_RTE_LIBRTE_EAL_LINUXAPP) += kni
endif
+ifeq ($(CONFIG_RTE_KDP_KMOD),y)
+DIRS-$(CONFIG_RTE_LIBRTE_EAL_LINUXAPP) += kdp
+endif
ifeq ($(CONFIG_RTE_LIBRTE_XEN_DOM0),y)
DIRS-$(CONFIG_RTE_LIBRTE_EAL_LINUXAPP) += xen_dom0
endif
diff --git a/lib/librte_eal/linuxapp/eal/Makefile b/lib/librte_eal/linuxapp/eal/Makefile
index 26eced5..ac72aea 100644
--- a/lib/librte_eal/linuxapp/eal/Makefile
+++ b/lib/librte_eal/linuxapp/eal/Makefile
@@ -1,6 +1,6 @@
# BSD LICENSE
#
-# Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
+# Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
@@ -116,6 +116,7 @@ CFLAGS_eal_thread.o += -Wno-return-type
endif
INC := rte_interrupts.h rte_kni_common.h rte_dom0_common.h
+INC += rte_kdp_common.h
SYMLINK-$(CONFIG_RTE_LIBRTE_EAL_LINUXAPP)-include/exec-env := \
$(addprefix include/exec-env/,$(INC))
diff --git a/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kdp_common.h b/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kdp_common.h
new file mode 100644
index 0000000..71f7dc9
--- /dev/null
+++ b/lib/librte_eal/linuxapp/eal/include/exec-env/rte_kdp_common.h
@@ -0,0 +1,138 @@
+/*-
+ * This file is provided under a dual BSD/LGPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GNU LESSER GENERAL PUBLIC LICENSE
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Contact Information:
+ * Intel Corporation
+ *
+ *
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#ifndef _RTE_KDP_COMMON_H_
+#define _RTE_KDP_COMMON_H_
+
+#ifdef __KERNEL__
+#include <linux/if.h>
+#endif
+
+/**
+ * KDP name is part of memzone name.
+ */
+#define RTE_KDP_NAMESIZE 32
+
+#ifndef RTE_CACHE_LINE_SIZE
+#define RTE_CACHE_LINE_SIZE 64 /**< Cache line size. */
+#endif
+
+/*
+ * Fifo struct mapped in a shared memory. It describes a circular buffer FIFO
+ * Write and read should wrap around. Fifo is empty when write == read
+ * Writing should never overwrite the read position
+ */
+struct rte_kdp_fifo {
+ volatile unsigned write; /**< Next position to be written*/
+ volatile unsigned read; /**< Next position to be read */
+ unsigned len; /**< Circular buffer length */
+ unsigned elem_size; /**< Pointer size - for 32/64 bit OS */
+ void * volatile buffer[0]; /**< The buffer contains mbuf pointers */
+};
+
+/*
+ * The kernel image of the rte_mbuf struct, with only the relevant fields.
+ * Padding is necessary to assure the offsets of these fields
+ */
+struct rte_kdp_mbuf {
+ void *buf_addr __attribute__((__aligned__(RTE_CACHE_LINE_SIZE)));
+ char pad0[10];
+ uint16_t data_off; /**< Start address of data in segment buffer. */
+ char pad1[4];
+ uint64_t ol_flags; /**< Offload features. */
+ char pad2[4];
+ uint32_t pkt_len; /**< Total pkt len: sum of all segment data_len. */
+ uint16_t data_len; /**< Amount of data in segment buffer. */
+
+ /* fields on second cache line */
+ char pad3[8] __attribute__((__aligned__(RTE_CACHE_LINE_SIZE)));
+ void *pool;
+ void *next;
+};
+
+/*
+ * Struct used to create a KDP device. Passed to the kernel in IOCTL call
+ */
+struct rte_kdp_device_info {
+ char name[RTE_KDP_NAMESIZE]; /**< Network device name for KDP */
+
+ phys_addr_t tx_phys;
+ phys_addr_t rx_phys;
+ phys_addr_t alloc_phys;
+ phys_addr_t free_phys;
+
+ /* mbuf mempool */
+ void * mbuf_va;
+ phys_addr_t mbuf_phys;
+
+ uint16_t group_id; /**< Group ID */
+ uint32_t core_id; /**< core ID to bind for kernel thread */
+
+ uint8_t force_bind : 1; /**< Flag for kernel thread binding */
+
+ /* mbuf size */
+ unsigned mbuf_size;
+};
+
+#define KDP_DEVICE "kdp"
+
+#define RTE_KDP_IOCTL_TEST _IOWR(0, 1, int)
+#define RTE_KDP_IOCTL_CREATE _IOWR(0, 2, struct rte_kdp_device_info)
+#define RTE_KDP_IOCTL_RELEASE _IOWR(0, 3, struct rte_kdp_device_info)
+
+#endif /* _RTE_KDP_COMMON_H_ */
diff --git a/lib/librte_eal/linuxapp/kdp/Makefile b/lib/librte_eal/linuxapp/kdp/Makefile
new file mode 100644
index 0000000..764f6a8
--- /dev/null
+++ b/lib/librte_eal/linuxapp/kdp/Makefile
@@ -0,0 +1,56 @@
+# BSD LICENSE
+#
+# Copyright(c) 2016 Intel Corporation. All rights reserved.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in
+# the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Intel Corporation nor the names of its
+# contributors may be used to endorse or promote products derived
+# from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# module name and path
+#
+MODULE = rte_kdp
+
+#
+# CFLAGS
+#
+MODULE_CFLAGS += -I$(SRCDIR) --param max-inline-insns-single=50
+MODULE_CFLAGS += -I$(RTE_OUTPUT)/include
+MODULE_CFLAGS += -include $(RTE_OUTPUT)/include/rte_config.h
+MODULE_CFLAGS += -Wall -Werror
+
+# this lib needs main eal
+DEPDIRS-y += lib/librte_eal/linuxapp/eal
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-y += kdp_misc.c
+SRCS-y += kdp_net.c
+
+include $(RTE_SDK)/mk/rte.module.mk
diff --git a/lib/librte_eal/linuxapp/kdp/compat.h b/lib/librte_eal/linuxapp/kdp/compat.h
new file mode 100644
index 0000000..103c094
--- /dev/null
+++ b/lib/librte_eal/linuxapp/kdp/compat.h
@@ -0,0 +1,29 @@
+/*
+ * Minimal wrappers to allow compiling kdp on older kernels.
+ */
+
+#ifndef RHEL_RELEASE_VERSION
+#define RHEL_RELEASE_VERSION(a, b) (((a) << 8) + (b))
+#endif
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 39) && \
+ (!(defined(RHEL_RELEASE_CODE) && \
+ RHEL_RELEASE_CODE >= RHEL_RELEASE_VERSION(6, 4)))
+
+#define kstrtoul strict_strtoul
+
+#endif /* < 2.6.39 */
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35)
+
+#define sk_sleep(s) (s)->sk_sleep
+
+#endif /* < 2.6.35 */
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,19,0)
+#define HAVE_IOV_ITER_MSGHDR
+#endif
+
+#if ( LINUX_VERSION_CODE < KERNEL_VERSION(4,1,0) )
+#define HAVE_KIOCB_MSG_PARAM
+#endif /* < 4.1.0 */
diff --git a/lib/librte_eal/linuxapp/kdp/kdp_dev.h b/lib/librte_eal/linuxapp/kdp/kdp_dev.h
new file mode 100644
index 0000000..a6f4bd4
--- /dev/null
+++ b/lib/librte_eal/linuxapp/kdp/kdp_dev.h
@@ -0,0 +1,85 @@
+/*-
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ * The full GNU General Public License is included in this distribution
+ * in the file called LICENSE.GPL.
+ *
+ * Contact Information:
+ * Intel Corporation
+ */
+
+#ifndef _KDP_DEV_H_
+#define _KDP_DEV_H_
+
+#include <exec-env/rte_kdp_common.h>
+
+/**
+ * A structure describing the private information for a kdp device.
+ */
+struct kdp_dev {
+ /* kdp list */
+ struct list_head list;
+
+ struct net_device_stats stats;
+ uint16_t group_id; /* Group ID of a group of KDP devices */
+ unsigned core_id; /* Core ID to bind */
+ char name[RTE_KDP_NAMESIZE]; /* Network device name */
+ struct task_struct *pthread;
+
+ /* wait queue for req/resp */
+ wait_queue_head_t wq;
+ struct mutex sync_lock;
+
+ /* kdp device */
+ struct net_device *net_dev;
+
+ /* queue for packets to be sent out */
+ void *tx_q;
+
+ /* queue for the packets received */
+ void *rx_q;
+
+ /* queue for the allocated mbufs those can be used to save sk buffs */
+ void *alloc_q;
+
+ /* free queue for the mbufs to be freed */
+ void *free_q;
+
+ void * sync_kva;
+ void *sync_va;
+
+ void *mbuf_kva;
+ void *mbuf_va;
+
+ /* mbuf size */
+ unsigned mbuf_size;
+};
+
+void kdp_net_rx(struct kdp_dev *kdp);
+void kdp_net_init(struct net_device *dev);
+void kdp_net_config_lo_mode(char *lo_str);
+
+#define KDP_ERR(args...) printk(KERN_DEBUG "KDP: Error: " args)
+#define KDP_PRINT(args...) printk(KERN_DEBUG "KDP: " args)
+
+#ifdef RTE_KDP_KO_DEBUG
+#define KDP_DBG(args...) printk(KERN_DEBUG "KDP: " args)
+#else
+#define KDP_DBG(args...)
+#endif
+
+#endif
diff --git a/lib/librte_eal/linuxapp/kdp/kdp_fifo.h b/lib/librte_eal/linuxapp/kdp/kdp_fifo.h
new file mode 100644
index 0000000..ea05d9e
--- /dev/null
+++ b/lib/librte_eal/linuxapp/kdp/kdp_fifo.h
@@ -0,0 +1,94 @@
+/*-
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ * The full GNU General Public License is included in this distribution
+ * in the file called LICENSE.GPL.
+ *
+ * Contact Information:
+ * Intel Corporation
+ */
+
+#ifndef _KDP_FIFO_H_
+#define _KDP_FIFO_H_
+
+#include <exec-env/rte_kdp_common.h>
+
+/**
+ * Adds num elements into the fifo. Return the number actually written
+ */
+static inline unsigned
+kdp_fifo_put(struct rte_kdp_fifo *fifo, void **data, unsigned num)
+{
+ unsigned i = 0;
+ unsigned fifo_write = fifo->write;
+ unsigned fifo_read = fifo->read;
+ unsigned new_write = fifo_write;
+
+ for (i = 0; i < num; i++) {
+ new_write = (new_write + 1) & (fifo->len - 1);
+
+ if (new_write == fifo_read)
+ break;
+ fifo->buffer[fifo_write] = data[i];
+ fifo_write = new_write;
+ }
+ fifo->write = fifo_write;
+
+ return i;
+}
+
+/**
+ * Get up to num elements from the fifo. Return the number actully read
+ */
+static inline unsigned
+kdp_fifo_get(struct rte_kdp_fifo *fifo, void **data, unsigned num)
+{
+ unsigned i = 0;
+ unsigned new_read = fifo->read;
+ unsigned fifo_write = fifo->write;
+
+ for (i = 0; i < num; i++) {
+ if (new_read == fifo_write)
+ break;
+
+ data[i] = fifo->buffer[new_read];
+ new_read = (new_read + 1) & (fifo->len - 1);
+ }
+ fifo->read = new_read;
+
+ return i;
+}
+
+/**
+ * Get the num of elements in the fifo
+ */
+static inline unsigned
+kdp_fifo_count(struct rte_kdp_fifo *fifo)
+{
+ return (fifo->len + fifo->write - fifo->read) & ( fifo->len - 1);
+}
+
+/**
+ * Get the num of available elements in the fifo
+ */
+static inline unsigned
+kdp_fifo_free_count(struct rte_kdp_fifo *fifo)
+{
+ return (fifo->read - fifo->write - 1) & (fifo->len - 1);
+}
+
+#endif /* _KDP_FIFO_H_ */
diff --git a/lib/librte_eal/linuxapp/kdp/kdp_misc.c b/lib/librte_eal/linuxapp/kdp/kdp_misc.c
new file mode 100644
index 0000000..2d58843
--- /dev/null
+++ b/lib/librte_eal/linuxapp/kdp/kdp_misc.c
@@ -0,0 +1,463 @@
+/*-
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation.
+ * The full GNU General Public License is included in this distribution
+ * in the file called LICENSE.GPL.
+ *
+ * Contact Information:
+ * Intel Corporation
+ */
+
+#include <linux/version.h>
+#include <linux/miscdevice.h>
+#include <linux/netdevice.h>
+#include <linux/pci.h>
+#include <linux/kthread.h>
+#include <net/netns/generic.h>
+
+#include "kdp_dev.h"
+
+#define KDP_RX_LOOP_NUM 1000
+#define KDP_DEV_IN_USE_BIT_NUM 0 /* Bit number for device in use */
+#define KDP_KTHREAD_RESCHEDULE_INTERVAL 5 /* us */
+
+static unsigned long device_in_use; /* device in use flag */
+static struct task_struct *kdp_kthread;
+static struct rw_semaphore kdp_list_lock;
+static struct list_head kdp_list_head;
+
+/* loopback mode */
+static char *lo_mode = NULL;
+
+/* Kernel thread mode */
+static char *kthread_mode = NULL;
+static unsigned multiple_kthread_on = 0;
+
+static int
+kdp_thread_single(void *data)
+{
+ struct kdp_dev *dev;
+ int j;
+
+ while (!kthread_should_stop()) {
+ down_read(&kdp_list_lock);
+ for (j = 0; j < KDP_RX_LOOP_NUM; j++) {
+ list_for_each_entry(dev, &kdp_list_head, list) {
+ kdp_net_rx(dev);
+ }
+ }
+ up_read(&kdp_list_lock);
+#ifdef RTE_KDP_PREEMPT_DEFAULT
+ /* reschedule out for a while */
+ schedule_timeout_interruptible(usecs_to_jiffies( \
+ KDP_KTHREAD_RESCHEDULE_INTERVAL));
+#endif
+ }
+
+ return 0;
+}
+
+static int
+kdp_thread_multiple(void *param)
+{
+ int j;
+ struct kdp_dev *dev = (struct kdp_dev *)param;
+
+ while (!kthread_should_stop()) {
+ for (j = 0; j < KDP_RX_LOOP_NUM; j++) {
+ kdp_net_rx(dev);
+ }
+#ifdef RTE_KDP_PREEMPT_DEFAULT
+ schedule_timeout_interruptible(usecs_to_jiffies( \
+ KDP_KTHREAD_RESCHEDULE_INTERVAL));
+#endif
+ }
+
+ return 0;
+}
+
+static int
+kdp_dev_remove(struct kdp_dev *dev)
+{
+ if (!dev)
+ return -ENODEV;
+
+ if (dev->net_dev) {
+ unregister_netdev(dev->net_dev);
+ free_netdev(dev->net_dev);
+ }
+
+ return 0;
+}
+
+static int
+kdp_check_param(struct kdp_dev *kdp, struct rte_kdp_device_info *dev)
+{
+ if (!kdp || !dev)
+ return -1;
+
+ /* Check if network name has been used */
+ if (!strncmp(kdp->name, dev->name, RTE_KDP_NAMESIZE)) {
+ KDP_ERR("KDP name %s duplicated\n", dev->name);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int
+kdp_ioctl_create(unsigned int ioctl_num, unsigned long ioctl_param)
+{
+ int ret;
+ struct rte_kdp_device_info dev_info;
+ struct net_device *net_dev = NULL;
+ struct kdp_dev *kdp, *dev, *n;
+
+ printk(KERN_INFO "KDP: Creating kdp...\n");
+ /* Check the buffer size, to avoid warning */
+ if (_IOC_SIZE(ioctl_num) > sizeof(dev_info))
+ return -EINVAL;
+
+ /* Copy kdp info from user space */
+ ret = copy_from_user(&dev_info, (void *)ioctl_param, sizeof(dev_info));
+ if (ret) {
+ KDP_ERR("copy_from_user in kdp_ioctl_create");
+ return -EIO;
+ }
+
+ /**
+ * Check if the cpu core id is valid for binding,
+ * for multiple kernel thread mode.
+ */
+ if (multiple_kthread_on && dev_info.force_bind &&
+ !cpu_online(dev_info.core_id)) {
+ KDP_ERR("cpu %u is not online\n", dev_info.core_id);
+ return -EINVAL;
+ }
+
+ /* Check if it has been created */
+ down_read(&kdp_list_lock);
+ list_for_each_entry_safe(dev, n, &kdp_list_head, list) {
+ if (kdp_check_param(dev, &dev_info) < 0) {
+ up_read(&kdp_list_lock);
+ return -EINVAL;
+ }
+ }
+ up_read(&kdp_list_lock);
+
+ net_dev = alloc_netdev(sizeof(struct kdp_dev), dev_info.name,
+#ifdef NET_NAME_UNKNOWN
+ NET_NAME_UNKNOWN,
+#endif
+ kdp_net_init);
+ if (net_dev == NULL) {
+ KDP_ERR("error allocating device \"%s\"\n", dev_info.name);
+ return -EBUSY;
+ }
+
+ kdp = netdev_priv(net_dev);
+
+ kdp->net_dev = net_dev;
+ kdp->group_id = dev_info.group_id;
+ kdp->core_id = dev_info.core_id;
+ strncpy(kdp->name, dev_info.name, RTE_KDP_NAMESIZE);
+
+ /* Translate user space info into kernel space info */
+ kdp->tx_q = phys_to_virt(dev_info.tx_phys);
+ kdp->rx_q = phys_to_virt(dev_info.rx_phys);
+ kdp->alloc_q = phys_to_virt(dev_info.alloc_phys);
+ kdp->free_q = phys_to_virt(dev_info.free_phys);
+
+ kdp->mbuf_kva = phys_to_virt(dev_info.mbuf_phys);
+ kdp->mbuf_va = dev_info.mbuf_va;
+
+ kdp->mbuf_size = dev_info.mbuf_size;
+
+ KDP_PRINT("tx_phys: 0x%016llx, tx_q addr: 0x%p\n",
+ (unsigned long long) dev_info.tx_phys, kdp->tx_q);
+ KDP_PRINT("rx_phys: 0x%016llx, rx_q addr: 0x%p\n",
+ (unsigned long long) dev_info.rx_phys, kdp->rx_q);
+ KDP_PRINT("alloc_phys: 0x%016llx, alloc_q addr: 0x%p\n",
+ (unsigned long long) dev_info.alloc_phys, kdp->alloc_q);
+ KDP_PRINT("free_phys: 0x%016llx, free_q addr: 0x%p\n",
+ (unsigned long long) dev_info.free_phys, kdp->free_q);
+ KDP_PRINT("mbuf_phys: 0x%016llx, mbuf_kva: 0x%p\n",
+ (unsigned long long) dev_info.mbuf_phys, kdp->mbuf_kva);
+ KDP_PRINT("mbuf_va: 0x%p\n", dev_info.mbuf_va);
+ KDP_PRINT("mbuf_size: %u\n", kdp->mbuf_size);
+
+ ret = register_netdev(net_dev);
+ if (ret) {
+ KDP_ERR("error %i registering device \"%s\"\n",
+ ret, dev_info.name);
+ kdp_dev_remove(kdp);
+ return -ENODEV;
+ }
+
+ /**
+ * Create a new kernel thread for multiple mode, set its core affinity,
+ * and finally wake it up.
+ */
+ if (multiple_kthread_on) {
+ kdp->pthread = kthread_create(kdp_thread_multiple,
+ (void *)kdp,
+ "kdp_%s", kdp->name);
+ if (IS_ERR(kdp->pthread)) {
+ kdp_dev_remove(kdp);
+ return -ECANCELED;
+ }
+ if (dev_info.force_bind)
+ kthread_bind(kdp->pthread, kdp->core_id);
+ wake_up_process(kdp->pthread);
+ }
+
+ down_write(&kdp_list_lock);
+ list_add(&kdp->list, &kdp_list_head);
+ up_write(&kdp_list_lock);
+
+ return 0;
+}
+
+static int
+kdp_ioctl_release(unsigned int ioctl_num, unsigned long ioctl_param)
+{
+ int ret = -EINVAL;
+ struct kdp_dev *dev, *n;
+ struct rte_kdp_device_info dev_info;
+
+ if (_IOC_SIZE(ioctl_num) > sizeof(dev_info))
+ return -EINVAL;
+
+ ret = copy_from_user(&dev_info, (void *)ioctl_param, sizeof(dev_info));
+ if (ret) {
+ KDP_ERR("copy_from_user in kdp_ioctl_release");
+ return -EIO;
+ }
+
+ /* Release the network device according to its name */
+ if (strlen(dev_info.name) == 0)
+ return ret;
+
+ down_write(&kdp_list_lock);
+ list_for_each_entry_safe(dev, n, &kdp_list_head, list) {
+ if (strncmp(dev->name, dev_info.name, RTE_KDP_NAMESIZE) != 0)
+ continue;
+
+ if (multiple_kthread_on && dev->pthread != NULL) {
+ kthread_stop(dev->pthread);
+ dev->pthread = NULL;
+ }
+
+ kdp_dev_remove(dev);
+ list_del(&dev->list);
+ ret = 0;
+ break;
+ }
+ up_write(&kdp_list_lock);
+ printk(KERN_INFO "KDP: %s release kdp named %s\n",
+ (ret == 0 ? "Successfully" : "Unsuccessfully"), dev_info.name);
+
+ return ret;
+}
+
+static int
+kdp_ioctl(struct inode *inode, unsigned int ioctl_num,
+ unsigned long ioctl_param)
+{
+ int ret = -EINVAL;
+
+ KDP_DBG("IOCTL num=0x%0x param=0x%0lx\n", ioctl_num, ioctl_param);
+
+ /*
+ * Switch according to the ioctl called
+ */
+ switch (_IOC_NR(ioctl_num)) {
+ case _IOC_NR(RTE_KDP_IOCTL_TEST):
+ /* For test only, not used */
+ break;
+ case _IOC_NR(RTE_KDP_IOCTL_CREATE):
+ ret = kdp_ioctl_create(ioctl_num, ioctl_param);
+ break;
+ case _IOC_NR(RTE_KDP_IOCTL_RELEASE):
+ ret = kdp_ioctl_release(ioctl_num, ioctl_param);
+ break;
+ default:
+ KDP_DBG("IOCTL default\n");
+ break;
+ }
+
+ return ret;
+}
+
+static int
+kdp_open(struct inode *inode, struct file *file)
+{
+ /* kdp device can be opened by one user only per netns */
+ if (test_and_set_bit(KDP_DEV_IN_USE_BIT_NUM, &device_in_use))
+ return -EBUSY;
+
+ /* Create kernel thread for single mode */
+ if (multiple_kthread_on == 0) {
+ KDP_PRINT("Single kernel thread for all KDP devices\n");
+ /* Create kernel thread for RX */
+ kdp_kthread = kthread_run(kdp_thread_single, NULL,
+ "kdp_single");
+ if (IS_ERR(kdp_kthread)) {
+ KDP_ERR("Unable to create kernel threaed\n");
+ return PTR_ERR(kdp_kthread);
+ }
+ } else
+ KDP_PRINT("Multiple kernel thread mode enabled\n");
+
+ KDP_PRINT("/dev/kdp opened\n");
+
+ return 0;
+}
+
+static int
+kdp_release(struct inode *inode, struct file *file)
+{
+ struct kdp_dev *dev, *n;
+
+ /* Stop kernel thread for single mode */
+ if (multiple_kthread_on == 0) {
+ /* Stop kernel thread */
+ kthread_stop(kdp_kthread);
+ kdp_kthread = NULL;
+ }
+
+ down_write(&kdp_list_lock);
+ list_for_each_entry_safe(dev, n, &kdp_list_head, list) {
+ /* Stop kernel thread for multiple mode */
+ if (multiple_kthread_on && dev->pthread != NULL) {
+ kthread_stop(dev->pthread);
+ dev->pthread = NULL;
+ }
+
+ kdp_dev_remove(dev);
+ list_del(&dev->list);
+ }
+ up_write(&kdp_list_lock);
+
+ /* Clear the bit of device in use */
+ clear_bit(KDP_DEV_IN_USE_BIT_NUM, &device_in_use);
+
+ KDP_PRINT("/dev/kdp closed\n");
+
+ return 0;
+}
+
+static int
+kdp_compat_ioctl(struct inode *inode, unsigned int ioctl_num,
+ unsigned long ioctl_param)
+{
+ /* 32 bits app on 64 bits OS to be supported later */
+ KDP_PRINT("Not implemented.\n");
+
+ return -EINVAL;
+}
+
+static struct file_operations kdp_fops = {
+ .owner = THIS_MODULE,
+ .open = kdp_open,
+ .release = kdp_release,
+ .unlocked_ioctl = (void *)kdp_ioctl,
+ .compat_ioctl = (void *)kdp_compat_ioctl,
+};
+
+static struct miscdevice kdp_misc = {
+ .minor = MISC_DYNAMIC_MINOR,
+ .name = KDP_DEVICE,
+ .fops = &kdp_fops,
+};
+
+static int __init
+kdp_parse_kthread_mode(void)
+{
+ if (!kthread_mode)
+ return 0;
+
+ if (strcmp(kthread_mode, "single") == 0)
+ return 0;
+ else if (strcmp(kthread_mode, "multiple") == 0)
+ multiple_kthread_on = 1;
+ else
+ return -1;
+
+ return 0;
+}
+
+static int __init
+kdp_init(void)
+{
+ int rc;
+
+ KDP_PRINT("######## DPDK kdp module loading ########\n");
+
+ if (kdp_parse_kthread_mode() < 0) {
+ KDP_ERR("Invalid parameter for kthread_mode\n");
+ return -EINVAL;
+ }
+
+ rc = misc_register(&kdp_misc);
+ if (rc != 0) {
+ KDP_ERR("Misc registration failed\n");
+ return rc;
+ }
+
+ /* Configure the lo mode according to the input parameter */
+ kdp_net_config_lo_mode(lo_mode);
+
+ /* Clear the bit of device in use */
+ clear_bit(KDP_DEV_IN_USE_BIT_NUM, &device_in_use);
+ init_rwsem(&kdp_list_lock);
+ INIT_LIST_HEAD(&kdp_list_head);
+
+ KDP_PRINT("######## DPDK kdp module loaded ########\n");
+
+ return 0;
+}
+module_init(kdp_init);
+
+static void __exit
+kdp_exit(void)
+{
+ misc_deregister(&kdp_misc);
+ KDP_PRINT("####### DPDK kdp module unloaded #######\n");
+}
+module_exit(kdp_exit);
+
+module_param(lo_mode, charp, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(lo_mode,
+"KDP loopback mode (default=lo_mode_none):\n"
+" lo_mode_none Kernel loopback disabled\n"
+" lo_mode_fifo Enable kernel loopback with fifo\n"
+" lo_mode_fifo_skb Enable kernel loopback with fifo and skb buffer\n"
+"\n"
+);
+
+module_param(kthread_mode, charp, S_IRUGO);
+MODULE_PARM_DESC(kthread_mode,
+"Kernel thread mode (default=single):\n"
+" single Single kernel thread mode enabled.\n"
+" multiple Multiple kernel thread mode enabled.\n"
+"\n"
+);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Intel Corporation");
+MODULE_DESCRIPTION("Kernel Module for managing kdp devices");
diff --git a/lib/librte_eal/linuxapp/kdp/kdp_net.c b/lib/librte_eal/linuxapp/kdp/kdp_net.c
new file mode 100644
index 0000000..376a6c7
--- /dev/null
+++ b/lib/librte_eal/linuxapp/kdp/kdp_net.c
@@ -0,0 +1,578 @@
+/*-
+ * GPL LICENSE SUMMARY
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ * The full GNU General Public License is included in this distribution
+ * in the file called LICENSE.GPL.
+ *
+ * Contact Information:
+ * Intel Corporation
+ */
+
+/*
+ * This code is inspired from the book "Linux Device Drivers" by
+ * Alessandro Rubini and Jonathan Corbet, published by O'Reilly & Associates
+ */
+
+#include <linux/version.h>
+#include <linux/etherdevice.h> /* eth_type_trans */
+
+#include "kdp_fifo.h"
+#include "kdp_dev.h"
+
+#define WD_TIMEOUT 5 /*jiffies */
+
+#define MBUF_BURST_SZ 32
+
+/* typedef for rx function */
+typedef void (*kdp_net_rx_t)(struct kdp_dev *kdp);
+
+/*
+ * Open and close
+ */
+static int
+kdp_net_open(struct net_device *dev)
+{
+ random_ether_addr(dev->dev_addr);
+ netif_start_queue(dev);
+
+ return 0;
+}
+
+static int
+kdp_net_release(struct net_device *dev)
+{
+ netif_stop_queue(dev); /* can't transmit any more */
+
+ return 0;
+}
+
+/*
+ * Configuration changes (passed on by ifconfig)
+ */
+static int
+kdp_net_config(struct net_device *dev, struct ifmap *map)
+{
+ if (dev->flags & IFF_UP) /* can't act on a running interface */
+ return -EBUSY;
+
+ /* ignore other fields */
+ return 0;
+}
+
+/*
+ * Transmit a packet (called by the kernel)
+ */
+static int
+kdp_net_tx(struct sk_buff *skb, struct net_device *dev)
+{
+ int len = 0;
+ unsigned ret;
+ struct kdp_dev *kdp = netdev_priv(dev);
+ struct rte_kdp_mbuf *pkt_kva = NULL;
+ struct rte_kdp_mbuf *pkt_va = NULL;
+
+ dev->trans_start = jiffies; /* save the timestamp */
+
+ /* Check if the length of skb is less than mbuf size */
+ if (skb->len > kdp->mbuf_size)
+ goto drop;
+
+ /**
+ * Check if it has at least one free entry in tx_q and
+ * one entry in alloc_q.
+ */
+ if (kdp_fifo_free_count(kdp->tx_q) == 0 ||
+ kdp_fifo_count(kdp->alloc_q) == 0) {
+ /**
+ * If no free entry in tx_q or no entry in alloc_q,
+ * drops skb and goes out.
+ */
+ goto drop;
+ }
+
+ /* dequeue a mbuf from alloc_q */
+ ret = kdp_fifo_get(kdp->alloc_q, (void **)&pkt_va, 1);
+ if (likely(ret == 1)) {
+ void *data_kva;
+
+ pkt_kva = (void *)pkt_va - kdp->mbuf_va + kdp->mbuf_kva;
+ data_kva = pkt_kva->buf_addr + pkt_kva->data_off - kdp->mbuf_va
+ + kdp->mbuf_kva;
+
+ len = skb->len;
+ memcpy(data_kva, skb->data, len);
+ if (unlikely(len < ETH_ZLEN)) {
+ memset(data_kva + len, 0, ETH_ZLEN - len);
+ len = ETH_ZLEN;
+ }
+ pkt_kva->pkt_len = len;
+ pkt_kva->data_len = len;
+
+ /* enqueue mbuf into tx_q */
+ ret = kdp_fifo_put(kdp->tx_q, (void **)&pkt_va, 1);
+ if (unlikely(ret != 1)) {
+ /* Failing should not happen */
+ KDP_ERR("Fail to enqueue mbuf into tx_q\n");
+ goto drop;
+ }
+ } else {
+ /* Failing should not happen */
+ KDP_ERR("Fail to dequeue mbuf from alloc_q\n");
+ goto drop;
+ }
+
+ /* Free skb and update statistics */
+ dev_kfree_skb(skb);
+ kdp->stats.tx_bytes += len;
+ kdp->stats.tx_packets++;
+
+ return NETDEV_TX_OK;
+
+drop:
+ /* Free skb and update statistics */
+ dev_kfree_skb(skb);
+ kdp->stats.tx_dropped++;
+
+ return NETDEV_TX_OK;
+}
+
+static int
+kdp_net_change_mtu(struct net_device *dev, int new_mtu)
+{
+ KDP_DBG("kdp_net_change_mtu new mtu %d to be set\n", new_mtu);
+
+ dev->mtu = new_mtu;
+
+ return 0;
+}
+
+/*
+ * Ioctl commands
+ */
+static int
+kdp_net_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
+{
+ KDP_DBG("kdp_net_ioctl %d\n",
+ ((struct kdp_dev *)netdev_priv(dev))->group_id);
+
+ return 0;
+}
+
+static void
+kdp_net_set_rx_mode(struct net_device *dev)
+{
+}
+
+/*
+ * Return statistics to the caller
+ */
+static struct net_device_stats *
+kdp_net_stats(struct net_device *dev)
+{
+ struct kdp_dev *kdp = netdev_priv(dev);
+ return &kdp->stats;
+}
+
+/*
+ * Deal with a transmit timeout.
+ */
+static void
+kdp_net_tx_timeout (struct net_device *dev)
+{
+ struct kdp_dev *kdp = netdev_priv(dev);
+
+ KDP_DBG("Transmit timeout at %ld, latency %ld\n", jiffies,
+ jiffies - dev->trans_start);
+
+ kdp->stats.tx_errors++;
+ netif_wake_queue(dev);
+ return;
+}
+
+/**
+ * kdp_net_set_mac - Change the Ethernet Address of the KDP NIC
+ * @netdev: network interface device structure
+ * @p: pointer to an address structure
+ *
+ * Returns 0 on success, negative on failure
+ **/
+static int kdp_net_set_mac(struct net_device *netdev, void *p)
+{
+ struct sockaddr *addr = p;
+ if (!is_valid_ether_addr((unsigned char *)(addr->sa_data)))
+ return -EADDRNOTAVAIL;
+ memcpy(netdev->dev_addr, addr->sa_data, netdev->addr_len);
+ return 0;
+}
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0))
+static int kdp_net_change_carrier(struct net_device *dev, bool new_carrier)
+{
+ if (new_carrier)
+ netif_carrier_on(dev);
+ else
+ netif_carrier_off(dev);
+ return 0;
+}
+#endif
+
+static const struct net_device_ops kdp_net_netdev_ops = {
+ .ndo_open = kdp_net_open,
+ .ndo_stop = kdp_net_release,
+ .ndo_set_config = kdp_net_config,
+ .ndo_start_xmit = kdp_net_tx,
+ .ndo_change_mtu = kdp_net_change_mtu,
+ .ndo_do_ioctl = kdp_net_ioctl,
+ .ndo_set_rx_mode = kdp_net_set_rx_mode,
+ .ndo_get_stats = kdp_net_stats,
+ .ndo_tx_timeout = kdp_net_tx_timeout,
+ .ndo_set_mac_address = kdp_net_set_mac,
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3, 9, 0))
+ .ndo_change_carrier = kdp_net_change_carrier,
+#endif
+};
+
+/*
+ * Fill the eth header
+ */
+static int
+kdp_net_header(struct sk_buff *skb, struct net_device *dev,
+ unsigned short type, const void *daddr,
+ const void *saddr, unsigned int len)
+{
+ struct ethhdr *eth = (struct ethhdr *) skb_push(skb, ETH_HLEN);
+
+ memcpy(eth->h_source, saddr ? saddr : dev->dev_addr, dev->addr_len);
+ memcpy(eth->h_dest, daddr ? daddr : dev->dev_addr, dev->addr_len);
+ eth->h_proto = htons(type);
+
+ return dev->hard_header_len;
+}
+
+/*
+ * Re-fill the eth header
+ */
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 1, 0))
+static int
+kdp_net_rebuild_header(struct sk_buff *skb)
+{
+ struct net_device *dev = skb->dev;
+ struct ethhdr *eth = (struct ethhdr *) skb->data;
+
+ memcpy(eth->h_source, dev->dev_addr, dev->addr_len);
+ memcpy(eth->h_dest, dev->dev_addr, dev->addr_len);
+
+ return 0;
+}
+#endif /* < 4.1.0 */
+
+static const struct header_ops kdp_net_header_ops = {
+ .create = kdp_net_header,
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 1, 0))
+ .rebuild = kdp_net_rebuild_header,
+#endif /* < 4.1.0 */
+ .cache = NULL, /* disable caching */
+};
+
+void
+kdp_net_init(struct net_device *dev)
+{
+ struct kdp_dev *kdp = netdev_priv(dev);
+
+ KDP_DBG("kdp_net_init\n");
+
+ init_waitqueue_head(&kdp->wq);
+ mutex_init(&kdp->sync_lock);
+
+ ether_setup(dev); /* assign some of the fields */
+ dev->netdev_ops = &kdp_net_netdev_ops;
+ dev->header_ops = &kdp_net_header_ops;
+ dev->watchdog_timeo = WD_TIMEOUT;
+}
+
+/*
+ * RX: normal working mode
+ */
+static void
+kdp_net_rx_normal(struct kdp_dev *kdp)
+{
+ unsigned ret;
+ uint32_t len;
+ unsigned i, num_rx, num_fq;
+ struct rte_kdp_mbuf *kva;
+ struct rte_kdp_mbuf *va[MBUF_BURST_SZ];
+ void *data_kva;
+
+ struct sk_buff *skb;
+ struct net_device *dev = kdp->net_dev;
+
+ /* Get the number of free entries in free_q */
+ num_fq = kdp_fifo_free_count(kdp->free_q);
+ if (num_fq == 0) {
+ /* No room on the free_q, bail out */
+ return;
+ }
+
+ /* Calculate the number of entries to dequeue from rx_q */
+ num_rx = min(num_fq, (unsigned)MBUF_BURST_SZ);
+
+ /* Burst dequeue from rx_q */
+ num_rx = kdp_fifo_get(kdp->rx_q, (void **)va, num_rx);
+ if (num_rx == 0)
+ return;
+
+ /* Transfer received packets to netif */
+ for (i = 0; i < num_rx; i++) {
+ kva = (void *)va[i] - kdp->mbuf_va + kdp->mbuf_kva;
+ len = kva->data_len;
+ data_kva = kva->buf_addr + kva->data_off - kdp->mbuf_va
+ + kdp->mbuf_kva;
+
+ skb = dev_alloc_skb(len + 2);
+ if (!skb) {
+ KDP_ERR("Out of mem, dropping pkts\n");
+ /* Update statistics */
+ kdp->stats.rx_dropped++;
+ }
+ else {
+ /* Align IP on 16B boundary */
+ skb_reserve(skb, 2);
+ memcpy(skb_put(skb, len), data_kva, len);
+ skb->dev = dev;
+ skb->protocol = eth_type_trans(skb, dev);
+ skb->ip_summed = CHECKSUM_UNNECESSARY;
+
+ /* Call netif interface */
+ netif_rx(skb);
+
+ /* Update statistics */
+ kdp->stats.rx_bytes += len;
+ kdp->stats.rx_packets++;
+ }
+ }
+
+ /* Burst enqueue mbufs into free_q */
+ ret = kdp_fifo_put(kdp->free_q, (void **)va, num_rx);
+ if (ret != num_rx)
+ /* Failing should not happen */
+ KDP_ERR("Fail to enqueue entries into free_q\n");
+}
+
+/*
+ * RX: loopback with enqueue/dequeue fifos.
+ */
+static void
+kdp_net_rx_lo_fifo(struct kdp_dev *kdp)
+{
+ unsigned ret;
+ uint32_t len;
+ unsigned i, num, num_rq, num_tq, num_aq, num_fq;
+ struct rte_kdp_mbuf *kva;
+ struct rte_kdp_mbuf *va[MBUF_BURST_SZ];
+ void * data_kva;
+
+ struct rte_kdp_mbuf *alloc_kva;
+ struct rte_kdp_mbuf *alloc_va[MBUF_BURST_SZ];
+ void *alloc_data_kva;
+
+ /* Get the number of entries in rx_q */
+ num_rq = kdp_fifo_count(kdp->rx_q);
+
+ /* Get the number of free entrie in tx_q */
+ num_tq = kdp_fifo_free_count(kdp->tx_q);
+
+ /* Get the number of entries in alloc_q */
+ num_aq = kdp_fifo_count(kdp->alloc_q);
+
+ /* Get the number of free entries in free_q */
+ num_fq = kdp_fifo_free_count(kdp->free_q);
+
+ /* Calculate the number of entries to be dequeued from rx_q */
+ num = min(num_rq, num_tq);
+ num = min(num, num_aq);
+ num = min(num, num_fq);
+ num = min(num, (unsigned)MBUF_BURST_SZ);
+
+ /* Return if no entry to dequeue from rx_q */
+ if (num == 0)
+ return;
+
+ /* Burst dequeue from rx_q */
+ ret = kdp_fifo_get(kdp->rx_q, (void **)va, num);
+ if (ret == 0)
+ return; /* Failing should not happen */
+
+ /* Dequeue entries from alloc_q */
+ ret = kdp_fifo_get(kdp->alloc_q, (void **)alloc_va, num);
+ if (ret) {
+ num = ret;
+ /* Copy mbufs */
+ for (i = 0; i < num; i++) {
+ kva = (void *)va[i] - kdp->mbuf_va + kdp->mbuf_kva;
+ len = kva->pkt_len;
+ data_kva = kva->buf_addr + kva->data_off -
+ kdp->mbuf_va + kdp->mbuf_kva;
+
+ alloc_kva = (void *)alloc_va[i] - kdp->mbuf_va +
+ kdp->mbuf_kva;
+ alloc_data_kva = alloc_kva->buf_addr +
+ alloc_kva->data_off - kdp->mbuf_va +
+ kdp->mbuf_kva;
+ memcpy(alloc_data_kva, data_kva, len);
+ alloc_kva->pkt_len = len;
+ alloc_kva->data_len = len;
+
+ kdp->stats.tx_bytes += len;
+ kdp->stats.rx_bytes += len;
+ }
+
+ /* Burst enqueue mbufs into tx_q */
+ ret = kdp_fifo_put(kdp->tx_q, (void **)alloc_va, num);
+ if (ret != num)
+ /* Failing should not happen */
+ KDP_ERR("Fail to enqueue mbufs into tx_q\n");
+ }
+
+ /* Burst enqueue mbufs into free_q */
+ ret = kdp_fifo_put(kdp->free_q, (void **)va, num);
+ if (ret != num)
+ /* Failing should not happen */
+ KDP_ERR("Fail to enqueue mbufs into free_q\n");
+
+ /**
+ * Update statistic, and enqueue/dequeue failure is impossible,
+ * as all queues are checked at first.
+ */
+ kdp->stats.tx_packets += num;
+ kdp->stats.rx_packets += num;
+}
+
+/*
+ * RX: loopback with enqueue/dequeue fifos and sk buffer copies.
+ */
+static void
+kdp_net_rx_lo_fifo_skb(struct kdp_dev *kdp)
+{
+ unsigned ret;
+ uint32_t len;
+ unsigned i, num_rq, num_fq, num;
+ struct rte_kdp_mbuf *kva;
+ struct rte_kdp_mbuf *va[MBUF_BURST_SZ];
+ void * data_kva;
+
+ struct sk_buff *skb;
+ struct net_device *dev = kdp->net_dev;
+
+ /* Get the number of entries in rx_q */
+ num_rq = kdp_fifo_count(kdp->rx_q);
+
+ /* Get the number of free entries in free_q */
+ num_fq = kdp_fifo_free_count(kdp->free_q);
+
+ /* Calculate the number of entries to dequeue from rx_q */
+ num = min(num_rq, num_fq);
+ num = min(num, (unsigned)MBUF_BURST_SZ);
+
+ /* Return if no entry to dequeue from rx_q */
+ if (num == 0)
+ return;
+
+ /* Burst dequeue mbufs from rx_q */
+ ret = kdp_fifo_get(kdp->rx_q, (void **)va, num);
+ if (ret == 0)
+ return;
+
+ /* Copy mbufs to sk buffer and then call tx interface */
+ for (i = 0; i < num; i++) {
+ kva = (void *)va[i] - kdp->mbuf_va + kdp->mbuf_kva;
+ len = kva->data_len;
+ data_kva = kva->buf_addr + kva->data_off - kdp->mbuf_va +
+ kdp->mbuf_kva;
+
+ skb = dev_alloc_skb(len + 2);
+ if (skb == NULL)
+ KDP_ERR("Out of mem, dropping pkts\n");
+ else {
+ /* Align IP on 16B boundary */
+ skb_reserve(skb, 2);
+ memcpy(skb_put(skb, len), data_kva, len);
+ skb->dev = dev;
+ skb->ip_summed = CHECKSUM_UNNECESSARY;
+ dev_kfree_skb(skb);
+ }
+
+ /* Simulate real usage, allocate/copy skb twice */
+ skb = dev_alloc_skb(len + 2);
+ if (skb == NULL) {
+ KDP_ERR("Out of mem, dropping pkts\n");
+ kdp->stats.rx_dropped++;
+ }
+ else {
+ /* Align IP on 16B boundary */
+ skb_reserve(skb, 2);
+ memcpy(skb_put(skb, len), data_kva, len);
+ skb->dev = dev;
+ skb->ip_summed = CHECKSUM_UNNECESSARY;
+
+ kdp->stats.rx_bytes += len;
+ kdp->stats.rx_packets++;
+
+ /* call tx interface */
+ kdp_net_tx(skb, dev);
+ }
+ }
+
+ /* enqueue all the mbufs from rx_q into free_q */
+ ret = kdp_fifo_put(kdp->free_q, (void **)&va, num);
+ if (ret != num)
+ /* Failing should not happen */
+ KDP_ERR("Fail to enqueue mbufs into free_q\n");
+}
+
+/* kdp rx function pointer, with default to normal rx */
+static kdp_net_rx_t kdp_net_rx_func = kdp_net_rx_normal;
+
+void
+kdp_net_config_lo_mode(char *lo_str)
+{
+ if (!lo_str) {
+ KDP_PRINT("loopback disabled");
+ return;
+ }
+
+ if (!strcmp(lo_str, "lo_mode_none"))
+ KDP_PRINT("loopback disabled");
+ else if (!strcmp(lo_str, "lo_mode_fifo")) {
+ KDP_PRINT("loopback mode=lo_mode_fifo enabled");
+ kdp_net_rx_func = kdp_net_rx_lo_fifo;
+ } else if (!strcmp(lo_str, "lo_mode_fifo_skb")) {
+ KDP_PRINT("loopback mode=lo_mode_fifo_skb enabled");
+ kdp_net_rx_func = kdp_net_rx_lo_fifo_skb;
+ } else
+ KDP_PRINT("Incognizant parameter, loopback disabled");
+}
+
+/* rx interface */
+void
+kdp_net_rx(struct kdp_dev *kdp)
+{
+ /**
+ * It doesn't need to check if it is NULL pointer,
+ * as it has a default value
+ */
+ (*kdp_net_rx_func)(kdp);
+}
--
2.5.0
^ permalink raw reply [flat|nested] 7+ messages in thread
* [dpdk-dev] [RFC 2/2] kdp: add virtual PMD for kernel slow data path communication
2016-01-22 16:00 [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 1/2] kdp: add kernel data path kernel module Ferruh Yigit
@ 2016-01-22 16:00 ` Ferruh Yigit
2016-01-22 16:31 ` [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Thomas Monjalon
2 siblings, 0 replies; 7+ messages in thread
From: Ferruh Yigit @ 2016-01-22 16:00 UTC (permalink / raw)
To: dev
This patch provides slow data path communication to the Linux kernel.
Patch is based on librte_kni, and heavily re-uses it.
The main difference is librte_kni library converted into a PMD, to
provide ease of use for applications.
Now any application can use slow path communication without any update
in application, because of existing eal support for virtual PMD.
PMD's rx_pkt_burst() get packets from FIFO, and tx_pkt_burst() puts
packet to the FIFO.
The corresponding Linux virtual network device driver code
also gets/puts packages from FIFO as they are coming from hardware.
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
config/common_linuxapp | 1 +
drivers/net/Makefile | 3 +-
drivers/net/kdp/Makefile | 60 +++++
drivers/net/kdp/rte_eth_kdp.c | 405 ++++++++++++++++++++++++++++++++
drivers/net/kdp/rte_kdp.c | 365 ++++++++++++++++++++++++++++
drivers/net/kdp/rte_kdp.h | 113 +++++++++
drivers/net/kdp/rte_kdp_fifo.h | 91 +++++++
drivers/net/kdp/rte_pmd_kdp_version.map | 4 +
lib/librte_eal/common/include/rte_log.h | 3 +-
mk/rte.app.mk | 3 +-
10 files changed, 1045 insertions(+), 3 deletions(-)
create mode 100644 drivers/net/kdp/Makefile
create mode 100644 drivers/net/kdp/rte_eth_kdp.c
create mode 100644 drivers/net/kdp/rte_kdp.c
create mode 100644 drivers/net/kdp/rte_kdp.h
create mode 100644 drivers/net/kdp/rte_kdp_fifo.h
create mode 100644 drivers/net/kdp/rte_pmd_kdp_version.map
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 73c91d8..b9dec0c 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -322,6 +322,7 @@ CONFIG_RTE_LIBRTE_PMD_NULL=y
#
# Compile KDP PMD
#
+CONFIG_RTE_LIBRTE_PMD_KDP=y
CONFIG_RTE_KDP_KMOD=y
CONFIG_RTE_KDP_PREEMPT_DEFAULT=y
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 6e4497e..0be06f5 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -1,6 +1,6 @@
# BSD LICENSE
#
-# Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
+# Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
@@ -51,6 +51,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_PMD_SZEDATA2) += szedata2
DIRS-$(CONFIG_RTE_LIBRTE_VIRTIO_PMD) += virtio
DIRS-$(CONFIG_RTE_LIBRTE_VMXNET3_PMD) += vmxnet3
DIRS-$(CONFIG_RTE_LIBRTE_PMD_XENVIRT) += xenvirt
+DIRS-$(CONFIG_RTE_LIBRTE_PMD_KDP) += kdp
include $(RTE_SDK)/mk/rte.sharelib.mk
include $(RTE_SDK)/mk/rte.subdir.mk
diff --git a/drivers/net/kdp/Makefile b/drivers/net/kdp/Makefile
new file mode 100644
index 0000000..ac44c0f
--- /dev/null
+++ b/drivers/net/kdp/Makefile
@@ -0,0 +1,60 @@
+# BSD LICENSE
+#
+# Copyright(c) 2016 Intel Corporation. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in
+# the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Intel Corporation nor the names of its
+# contributors may be used to endorse or promote products derived
+# from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_pmd_kdp.a
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+EXPORT_MAP := rte_pmd_kdp_version.map
+
+LIBABIVER := 1
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-$(CONFIG_RTE_LIBRTE_PMD_KDP) += rte_eth_kdp.c
+SRCS-$(CONFIG_RTE_LIBRTE_PMD_KDP) += rte_kdp.c
+
+#
+# Export include files
+#
+SYMLINK-y-include +=
+
+# this lib depends upon:
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KDP) += lib/librte_mbuf
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KDP) += lib/librte_ether
+
+include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/net/kdp/rte_eth_kdp.c b/drivers/net/kdp/rte_eth_kdp.c
new file mode 100644
index 0000000..463bac8
--- /dev/null
+++ b/drivers/net/kdp/rte_eth_kdp.c
@@ -0,0 +1,405 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <rte_mbuf.h>
+#include <rte_ethdev.h>
+#include <rte_malloc.h>
+#include <rte_memcpy.h>
+#include <rte_dev.h>
+#include <rte_kvargs.h>
+#include <rte_spinlock.h>
+#include "rte_kdp.h"
+
+#define MAX_PACKET_SZ 2048
+
+struct kdp_queue {
+ struct pmd_internals *internals;
+ struct rte_mempool *mb_pool;
+
+ uint64_t rx_pkts;
+ uint64_t rx_bytes;
+ uint64_t rx_err_pkts;
+ uint64_t tx_pkts;
+ uint64_t tx_bytes;
+ uint64_t tx_err_pkts;
+};
+
+struct pmd_internals {
+ struct rte_kdp *kdp;
+
+ struct kdp_queue rx_queues[RTE_MAX_QUEUES_PER_PORT];
+ struct kdp_queue tx_queues[RTE_MAX_QUEUES_PER_PORT];
+};
+
+static struct ether_addr eth_addr = { .addr_bytes = {0} };
+static const char *drivername = "KDP PMD";
+static struct rte_eth_link pmd_link = {
+ .link_speed = 10000,
+ .link_duplex = ETH_LINK_FULL_DUPLEX,
+ .link_status = 0
+};
+
+static uint16_t
+eth_kdp_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
+{
+ struct kdp_queue *kdp_q = q;
+ struct pmd_internals *internals = kdp_q->internals;
+ uint16_t nb_pkts;
+
+ nb_pkts = rte_kdp_rx_burst(internals->kdp, bufs, nb_bufs);
+
+ kdp_q->rx_pkts += nb_pkts;
+ kdp_q->rx_err_pkts += nb_bufs - nb_pkts;
+
+ return nb_pkts;
+}
+
+static uint16_t
+eth_kdp_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
+{
+ struct kdp_queue *kdp_q = q;
+ struct pmd_internals *internals = kdp_q->internals;
+ uint16_t nb_pkts;
+
+ nb_pkts = rte_kdp_tx_burst(internals->kdp, bufs, nb_bufs);
+
+ kdp_q->tx_pkts += nb_pkts;
+ kdp_q->tx_err_pkts += nb_bufs - nb_pkts;
+
+ return nb_pkts;
+}
+
+static int
+kdp_start(struct rte_eth_dev *dev)
+{
+ struct pmd_internals *internals = dev->data->dev_private;
+ struct rte_kdp_conf conf;
+ uint16_t port_id = dev->data->port_id;
+ int ret;
+
+ snprintf(conf.name, RTE_KDP_NAMESIZE, "kdp%u", port_id);
+ conf.force_bind = 0;
+ conf.group_id = port_id;
+ conf.mbuf_size = MAX_PACKET_SZ;
+
+ ret = rte_kdp_start(internals->kdp, internals->rx_queues[0].mb_pool,
+ &conf);
+ if (ret)
+ RTE_LOG(INFO, PMD, "Fail to create kdp for port: %d\n",
+ port_id);
+ return ret;
+}
+
+static int
+eth_dev_start(struct rte_eth_dev *dev)
+{
+ int ret;
+
+ ret = kdp_start(dev);
+ if (ret)
+ return -1;
+
+ dev->data->dev_link.link_status = 1;
+ return 0;
+}
+
+static void
+eth_dev_stop(struct rte_eth_dev *dev)
+{
+ struct pmd_internals *internals = dev->data->dev_private;
+
+ rte_kdp_release(internals->kdp);
+ dev->data->dev_link.link_status = 0;
+}
+
+static void
+eth_dev_close(struct rte_eth_dev *dev __rte_unused)
+{
+ rte_kdp_close();
+}
+
+static int
+eth_dev_configure(struct rte_eth_dev *dev __rte_unused)
+{
+ return 0;
+}
+
+static void
+eth_dev_info(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
+{
+ struct rte_eth_dev_data *data = dev->data;
+
+ dev_info->driver_name = data->drv_name;
+ dev_info->max_mac_addrs = 1;
+ dev_info->max_rx_pktlen = (uint32_t)-1;
+ dev_info->max_rx_queues = data->nb_rx_queues;
+ dev_info->max_tx_queues = data->nb_tx_queues;
+ dev_info->min_rx_bufsize = 0;
+ dev_info->pci_dev = NULL;
+}
+
+static int
+eth_rx_queue_setup(struct rte_eth_dev *dev,
+ uint16_t rx_queue_id __rte_unused,
+ uint16_t nb_rx_desc __rte_unused,
+ unsigned int socket_id __rte_unused,
+ const struct rte_eth_rxconf *rx_conf __rte_unused,
+ struct rte_mempool *mb_pool)
+{
+ struct pmd_internals *internals = dev->data->dev_private;
+ struct kdp_queue *q;
+
+ q = &internals->rx_queues[rx_queue_id];
+ q->internals = internals;
+ q->mb_pool = mb_pool;
+
+ dev->data->rx_queues[rx_queue_id] = q;
+
+ return 0;
+}
+
+static int
+eth_tx_queue_setup(struct rte_eth_dev *dev,
+ uint16_t tx_queue_id,
+ uint16_t nb_tx_desc __rte_unused,
+ unsigned int socket_id __rte_unused,
+ const struct rte_eth_txconf *tx_conf __rte_unused)
+{
+ struct pmd_internals *internals = dev->data->dev_private;
+ struct kdp_queue *q;
+
+ q = &internals->tx_queues[tx_queue_id];
+ q->internals = internals;
+
+ dev->data->tx_queues[tx_queue_id] = q;
+
+ return 0;
+}
+
+static void
+eth_queue_release(void *q __rte_unused)
+{
+}
+
+static int
+eth_link_update(struct rte_eth_dev *dev __rte_unused,
+ int wait_to_complete __rte_unused)
+{
+ return 0;
+}
+
+static void
+eth_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
+{
+ unsigned i, num_stats;
+ unsigned long rx_packets_total = 0, rx_bytes_total = 0;
+ unsigned long tx_packets_total = 0, tx_bytes_total = 0;
+ unsigned long tx_packets_err_total = 0;
+ struct rte_eth_dev_data *data = dev->data;
+ struct kdp_queue *q;
+
+ num_stats = RTE_MIN((unsigned)RTE_ETHDEV_QUEUE_STAT_CNTRS,
+ data->nb_rx_queues);
+ for (i = 0; i < num_stats; i++) {
+ q = data->rx_queues[i];
+ stats->q_ipackets[i] = q->rx_pkts;
+ stats->q_ibytes[i] = q->rx_bytes;
+ rx_packets_total += stats->q_ipackets[i];
+ rx_bytes_total += stats->q_ibytes[i];
+ }
+
+ num_stats = RTE_MIN((unsigned)RTE_ETHDEV_QUEUE_STAT_CNTRS,
+ data->nb_tx_queues);
+ for (i = 0; i < num_stats; i++) {
+ q = data->tx_queues[i];
+ stats->q_opackets[i] = q->tx_pkts;
+ stats->q_obytes[i] = q->tx_bytes;
+ stats->q_errors[i] = q->tx_err_pkts;
+ tx_packets_total += stats->q_opackets[i];
+ tx_bytes_total += stats->q_obytes[i];
+ tx_packets_err_total += stats->q_errors[i];
+ }
+
+ stats->ipackets = rx_packets_total;
+ stats->ibytes = rx_bytes_total;
+ stats->opackets = tx_packets_total;
+ stats->obytes = tx_bytes_total;
+ stats->oerrors = tx_packets_err_total;
+}
+
+static void
+eth_stats_reset(struct rte_eth_dev *dev)
+{
+ unsigned i;
+ struct rte_eth_dev_data *data = dev->data;
+ struct kdp_queue *q;
+
+ for (i = 0; i < data->nb_rx_queues; i++) {
+ q = data->rx_queues[i];
+ q->rx_pkts = 0;
+ q->rx_bytes = 0;
+ }
+ for (i = 0; i < data->nb_tx_queues; i++) {
+ q = data->rx_queues[i];
+ q->tx_pkts = 0;
+ q->tx_bytes = 0;
+ q->tx_err_pkts = 0;
+ }
+}
+
+static const struct eth_dev_ops ops = {
+ .dev_start = eth_dev_start,
+ .dev_stop = eth_dev_stop,
+ .dev_close = eth_dev_close,
+ .dev_configure = eth_dev_configure,
+ .dev_infos_get = eth_dev_info,
+ .rx_queue_setup = eth_rx_queue_setup,
+ .tx_queue_setup = eth_tx_queue_setup,
+ .rx_queue_release = eth_queue_release,
+ .tx_queue_release = eth_queue_release,
+ .link_update = eth_link_update,
+ .stats_get = eth_stats_get,
+ .stats_reset = eth_stats_reset,
+};
+
+static struct rte_eth_dev *
+eth_dev_kdp_create(const char *name, unsigned numa_node)
+{
+ uint16_t nb_rx_queues = 1;
+ uint16_t nb_tx_queues = 1;
+ struct rte_eth_dev_data *data = NULL;
+ struct pmd_internals *internals = NULL;
+ struct rte_eth_dev *eth_dev = NULL;
+
+ if (name == NULL)
+ return NULL;
+
+ RTE_LOG(INFO, PMD, "Creating kdp ethdev on numa socket %u\n",
+ numa_node);
+
+ /*
+ * now do all data allocation - for eth_dev structure
+ * and internal (private) data
+ */
+ internals = rte_zmalloc_socket(name, sizeof(*internals), 0, numa_node);
+ if (internals == NULL)
+ goto error;
+
+ /* reserve an ethdev entry */
+ eth_dev = rte_eth_dev_allocate(name, RTE_ETH_DEV_VIRTUAL);
+ if (eth_dev == NULL)
+ goto error;
+
+ data = eth_dev->data;
+ data->dev_private = internals;
+ data->nb_rx_queues = nb_rx_queues;
+ data->nb_tx_queues = nb_tx_queues;
+ data->dev_link = pmd_link;
+ data->mac_addrs = ð_addr;
+
+ eth_dev->dev_ops = &ops;
+ eth_dev->driver = NULL;
+
+ data->dev_flags = RTE_ETH_DEV_DETACHABLE;
+ data->kdrv = RTE_KDRV_NONE;
+ data->drv_name = drivername;
+ data->numa_node = numa_node;
+
+ eth_dev->rx_pkt_burst = eth_kdp_rx;
+ eth_dev->tx_pkt_burst = eth_kdp_tx;
+
+ return eth_dev;
+
+error:
+ rte_free(internals);
+
+ return NULL;
+}
+
+static int
+rte_pmd_kdp_devinit(const char *name, const char *params __rte_unused)
+{
+ struct rte_eth_dev *eth_dev = NULL;
+ struct pmd_internals *internals;
+ struct rte_kdp *kdp;
+
+ RTE_LOG(INFO, PMD, "Initializing eth_kdp for %s\n", name);
+
+ eth_dev = eth_dev_kdp_create(name, rte_socket_id());
+ if (eth_dev == NULL)
+ return -1;
+
+ kdp = rte_kdp_init(eth_dev->data->port_id);
+ if (kdp == NULL) {
+ rte_eth_dev_release_port(eth_dev);
+ rte_free(eth_dev->data->dev_private);
+
+ /* Not return error to prevent panic in rte_eal_init() */
+ return 0;
+ }
+
+ internals = eth_dev->data->dev_private;
+ internals->kdp = kdp;
+ return 0;
+}
+
+static int
+rte_pmd_kdp_devuninit(const char *name)
+{
+ struct rte_eth_dev *eth_dev = NULL;
+
+ if (name == NULL)
+ return -EINVAL;
+
+ RTE_LOG(INFO, PMD, "Un-Initializing eth_kdp for %s\n", name);
+
+ /* find the ethdev entry */
+ eth_dev = rte_eth_dev_allocated(name);
+ if (eth_dev == NULL)
+ return -1;
+
+ rte_eth_dev_release_port(eth_dev);
+
+ rte_free(eth_dev->data->dev_private);
+
+ return 0;
+}
+
+static struct rte_driver pmd_kdp_drv = {
+ .name = "eth_kdp",
+ .type = PMD_VDEV,
+ .init = rte_pmd_kdp_devinit,
+ .uninit = rte_pmd_kdp_devuninit,
+};
+
+PMD_REGISTER_DRIVER(pmd_kdp_drv);
diff --git a/drivers/net/kdp/rte_kdp.c b/drivers/net/kdp/rte_kdp.c
new file mode 100644
index 0000000..3d0a184
--- /dev/null
+++ b/drivers/net/kdp/rte_kdp.c
@@ -0,0 +1,365 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef RTE_EXEC_ENV_LINUXAPP
+#error "KDP is not supported"
+#endif
+
+#include <fcntl.h>
+#include <unistd.h>
+
+#include <sys/ioctl.h>
+
+#include <rte_spinlock.h>
+#include <rte_ethdev.h>
+#include <rte_malloc.h>
+#include <rte_memzone.h>
+
+#include "rte_kdp.h"
+#include "rte_kdp_fifo.h"
+
+#define MAX_MBUF_BURST_NUM 32
+
+/* Maximum number of ring entries */
+#define KDP_FIFO_COUNT_MAX 1024
+#define KDP_FIFO_SIZE (KDP_FIFO_COUNT_MAX * sizeof(void *) + \
+ sizeof(struct rte_kdp_fifo))
+
+#define KDP_MEM_CHECK(cond) do { if (cond) goto kdp_fail; } while (0)
+
+static volatile int kdp_fd = -1;
+
+static const struct rte_memzone *
+kdp_memzone_reserve(const char *name, size_t len, int socket_id,
+ unsigned flags)
+{
+ const struct rte_memzone *mz = rte_memzone_lookup(name);
+
+ if (mz == NULL)
+ mz = rte_memzone_reserve(name, len, socket_id, flags);
+
+ return mz;
+}
+
+static int
+slot_init(struct rte_kdp_memzone_slot *slot)
+{
+#define OBJNAMSIZ 32
+ char obj_name[OBJNAMSIZ];
+ const struct rte_memzone *mz;
+
+ /* TX RING */
+ snprintf(obj_name, OBJNAMSIZ, "kdp_tx_%d", slot->id);
+ mz = kdp_memzone_reserve(obj_name, KDP_FIFO_SIZE, SOCKET_ID_ANY, 0);
+ KDP_MEM_CHECK(mz == NULL);
+ slot->m_tx_q = mz;
+
+ /* RX RING */
+ snprintf(obj_name, OBJNAMSIZ, "kdp_rx_%d", slot->id);
+ mz = kdp_memzone_reserve(obj_name, KDP_FIFO_SIZE, SOCKET_ID_ANY, 0);
+ KDP_MEM_CHECK(mz == NULL);
+ slot->m_rx_q = mz;
+
+ /* ALLOC RING */
+ snprintf(obj_name, OBJNAMSIZ, "kdp_alloc_%d", slot->id);
+ mz = kdp_memzone_reserve(obj_name, KDP_FIFO_SIZE, SOCKET_ID_ANY, 0);
+ KDP_MEM_CHECK(mz == NULL);
+ slot->m_alloc_q = mz;
+
+ /* FREE RING */
+ snprintf(obj_name, OBJNAMSIZ, "kdp_free_%d", slot->id);
+ mz = kdp_memzone_reserve(obj_name, KDP_FIFO_SIZE, SOCKET_ID_ANY, 0);
+ KDP_MEM_CHECK(mz == NULL);
+ slot->m_free_q = mz;
+
+ return 0;
+
+kdp_fail:
+ return -1;
+
+}
+
+static void
+ring_init(struct rte_kdp *kdp)
+{
+ struct rte_kdp_memzone_slot *slot = kdp->slot;
+ const struct rte_memzone *mz;
+
+ /* TX RING */
+ mz = slot->m_tx_q;
+ kdp->tx_q = mz->addr;
+ kdp_fifo_init(kdp->tx_q, KDP_FIFO_COUNT_MAX);
+
+ /* RX RING */
+ mz = slot->m_rx_q;
+ kdp->rx_q = mz->addr;
+ kdp_fifo_init(kdp->rx_q, KDP_FIFO_COUNT_MAX);
+
+ /* ALLOC RING */
+ mz = slot->m_alloc_q;
+ kdp->alloc_q = mz->addr;
+ kdp_fifo_init(kdp->alloc_q, KDP_FIFO_COUNT_MAX);
+
+ /* FREE RING */
+ mz = slot->m_free_q;
+ kdp->free_q = mz->addr;
+ kdp_fifo_init(kdp->free_q, KDP_FIFO_COUNT_MAX);
+}
+
+/* Shall be called before any allocation happens */
+struct rte_kdp *
+rte_kdp_init(uint16_t port_id)
+{
+ struct rte_kdp_memzone_slot *slot = NULL;
+ struct rte_kdp *kdp = NULL;
+ int ret;
+
+ /* Check FD and open */
+ if (kdp_fd < 0) {
+ kdp_fd = open("/dev/kdp", O_RDWR);
+ if (kdp_fd < 0) {
+ RTE_LOG(ERR, KDP,"Can not open /dev/kdp\n");
+ return NULL;
+ }
+ }
+
+ slot = rte_malloc(NULL, sizeof(struct rte_kdp_memzone_slot), 0);
+ KDP_MEM_CHECK(slot == NULL);
+ slot->id = port_id;
+
+ kdp = rte_malloc(NULL, sizeof(struct rte_kdp), 0);
+ KDP_MEM_CHECK(kdp == NULL);
+ kdp->slot = slot;
+
+ ret = slot_init(slot);
+ KDP_MEM_CHECK(ret < 0);
+
+ ring_init(kdp);
+
+ return kdp;
+
+kdp_fail:
+ rte_free(slot);
+ rte_free(kdp);
+ RTE_LOG(ERR, KDP, "Unable to allocate memory\n");
+ return NULL;
+}
+
+static void
+kdp_allocate_mbufs(struct rte_kdp *kdp)
+{
+ int i, ret;
+ struct rte_mbuf *pkts[MAX_MBUF_BURST_NUM];
+
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, pool) !=
+ offsetof(struct rte_kdp_mbuf, pool));
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, buf_addr) !=
+ offsetof(struct rte_kdp_mbuf, buf_addr));
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, next) !=
+ offsetof(struct rte_kdp_mbuf, next));
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, data_off) !=
+ offsetof(struct rte_kdp_mbuf, data_off));
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, data_len) !=
+ offsetof(struct rte_kdp_mbuf, data_len));
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, pkt_len) !=
+ offsetof(struct rte_kdp_mbuf, pkt_len));
+ RTE_BUILD_BUG_ON(offsetof(struct rte_mbuf, ol_flags) !=
+ offsetof(struct rte_kdp_mbuf, ol_flags));
+
+ /* Check if pktmbuf pool has been configured */
+ if (kdp->pktmbuf_pool == NULL) {
+ RTE_LOG(ERR, KDP, "No valid mempool for allocating mbufs\n");
+ return;
+ }
+
+ for (i = 0; i < MAX_MBUF_BURST_NUM; i++) {
+ pkts[i] = rte_pktmbuf_alloc(kdp->pktmbuf_pool);
+ if (unlikely(pkts[i] == NULL)) {
+ /* Out of memory */
+ RTE_LOG(ERR, KDP, "Out of memory\n");
+ break;
+ }
+ }
+
+ /* No pkt mbuf alocated */
+ if (i <= 0)
+ return;
+
+ ret = kdp_fifo_put(kdp->alloc_q, (void **)pkts, i);
+
+ /* Check if any mbufs not put into alloc_q, and then free them */
+ if (ret >= 0 && ret < i && ret < MAX_MBUF_BURST_NUM) {
+ int j;
+
+ for (j = ret; j < i; j++)
+ rte_pktmbuf_free(pkts[j]);
+ }
+}
+
+int
+rte_kdp_start(struct rte_kdp *kdp, struct rte_mempool *pktmbuf_pool,
+ const struct rte_kdp_conf *conf)
+{
+ struct rte_kdp_memzone_slot *slot = kdp->slot;
+ struct rte_kdp_device_info dev_info;
+ char mz_name[RTE_MEMZONE_NAMESIZE];
+ const struct rte_memzone *mz;
+ int ret;
+
+ if (!kdp || !pktmbuf_pool || !conf || !conf->name[0])
+ return -1;
+
+ snprintf(kdp->name, RTE_KDP_NAMESIZE, "%s", conf->name);
+ kdp->pktmbuf_pool = pktmbuf_pool;
+ kdp->group_id = conf->group_id;
+
+ memset(&dev_info, 0, sizeof(dev_info));
+ dev_info.core_id = conf->core_id;
+ dev_info.force_bind = conf->force_bind;
+ dev_info.group_id = conf->group_id;
+ dev_info.mbuf_size = conf->mbuf_size;
+ snprintf(dev_info.name, RTE_KDP_NAMESIZE, "%s", conf->name);
+
+ dev_info.tx_phys = slot->m_tx_q->phys_addr;
+ dev_info.rx_phys = slot->m_rx_q->phys_addr;
+ dev_info.alloc_phys = slot->m_alloc_q->phys_addr;
+ dev_info.free_phys = slot->m_free_q->phys_addr;
+
+ /* MBUF mempool */
+ snprintf(mz_name, sizeof(mz_name), RTE_MEMPOOL_OBJ_NAME,
+ pktmbuf_pool->name);
+ mz = rte_memzone_lookup(mz_name);
+ KDP_MEM_CHECK(mz == NULL);
+ dev_info.mbuf_va = mz->addr;
+ dev_info.mbuf_phys = mz->phys_addr;
+
+ ret = ioctl(kdp_fd, RTE_KDP_IOCTL_CREATE, &dev_info);
+ KDP_MEM_CHECK(ret < 0);
+
+ kdp->in_use = 1;
+
+ /* Allocate mbufs and then put them into alloc_q */
+ kdp_allocate_mbufs(kdp);
+
+ return 0;
+
+kdp_fail:
+ return -1;
+}
+
+static void
+kdp_free_mbufs(struct rte_kdp *kdp)
+{
+ int i, ret;
+ struct rte_mbuf *pkts[MAX_MBUF_BURST_NUM];
+
+ ret = kdp_fifo_get(kdp->free_q, (void **)pkts, MAX_MBUF_BURST_NUM);
+ if (likely(ret > 0)) {
+ for (i = 0; i < ret; i++)
+ rte_pktmbuf_free(pkts[i]);
+ }
+}
+
+unsigned
+rte_kdp_tx_burst(struct rte_kdp *kdp, struct rte_mbuf **mbufs, unsigned num)
+{
+ unsigned ret = kdp_fifo_put(kdp->rx_q, (void **)mbufs, num);
+
+ /* Get mbufs from free_q and then free them */
+ kdp_free_mbufs(kdp);
+
+ return ret;
+}
+
+unsigned
+rte_kdp_rx_burst(struct rte_kdp *kdp, struct rte_mbuf **mbufs, unsigned num)
+{
+ unsigned ret = kdp_fifo_get(kdp->tx_q, (void **)mbufs, num);
+
+ /* If buffers removed, allocate mbufs and then put them into alloc_q */
+ if (ret)
+ kdp_allocate_mbufs(kdp);
+
+ return ret;
+}
+
+static void
+kdp_free_fifo(struct rte_kdp_fifo *fifo)
+{
+ int ret;
+ struct rte_mbuf *pkt;
+
+ do {
+ ret = kdp_fifo_get(fifo, (void **)&pkt, 1);
+ if (ret)
+ rte_pktmbuf_free(pkt);
+ } while (ret);
+}
+
+int
+rte_kdp_release(struct rte_kdp *kdp)
+{
+ struct rte_kdp_device_info dev_info;
+
+ if (!kdp || !kdp->in_use)
+ return -1;
+
+ snprintf(dev_info.name, sizeof(dev_info.name), "%s", kdp->name);
+ if (ioctl(kdp_fd, RTE_KDP_IOCTL_RELEASE, &dev_info) < 0) {
+ RTE_LOG(ERR, KDP, "Fail to release kdp device\n");
+ return -1;
+ }
+
+ /* mbufs in all fifo should be released, except request/response */
+ kdp_free_fifo(kdp->tx_q);
+ kdp_free_fifo(kdp->rx_q);
+ kdp_free_fifo(kdp->alloc_q);
+ kdp_free_fifo(kdp->free_q);
+
+ rte_free(kdp->slot);
+
+ /* Memset the KDP struct */
+ memset(kdp, 0, sizeof(struct rte_kdp));
+
+ return 0;
+}
+
+void
+rte_kdp_close(void)
+{
+ if (kdp_fd < 0)
+ return;
+
+ close(kdp_fd);
+ kdp_fd = -1;
+}
diff --git a/drivers/net/kdp/rte_kdp.h b/drivers/net/kdp/rte_kdp.h
new file mode 100644
index 0000000..28f959e
--- /dev/null
+++ b/drivers/net/kdp/rte_kdp.h
@@ -0,0 +1,113 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef _RTE_KDP_H_
+#define _RTE_KDP_H_
+
+#include <rte_memory.h>
+#include <rte_mempool.h>
+
+#include <exec-env/rte_kdp_common.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * KDP memzone pool slot
+ */
+struct rte_kdp_memzone_slot {
+ uint32_t id;
+
+ /* Memzones */
+ const struct rte_memzone *m_tx_q; /**< TX queue */
+ const struct rte_memzone *m_rx_q; /**< RX queue */
+ const struct rte_memzone *m_alloc_q; /**< Allocated mbufs queue */
+ const struct rte_memzone *m_free_q; /**< To be freed mbufs queue */
+};
+
+/**
+ * KDP context
+ */
+struct rte_kdp {
+ char name[RTE_KDP_NAMESIZE]; /**< KDP interface name */
+ struct rte_mempool *pktmbuf_pool; /**< pkt mbuf mempool */
+ struct rte_kdp_memzone_slot *slot;
+ uint16_t group_id; /**< Group ID of KDP devices */
+
+ struct rte_kdp_fifo *tx_q; /**< TX queue */
+ struct rte_kdp_fifo *rx_q; /**< RX queue */
+ struct rte_kdp_fifo *alloc_q; /**< Allocated mbufs queue */
+ struct rte_kdp_fifo *free_q; /**< To be freed mbufs queue */
+
+ uint8_t in_use; /**< kdp in use */
+
+};
+
+/**
+ * Structure for configuring KDP device.
+ */
+struct rte_kdp_conf {
+ /*
+ * KDP name which will be used in relevant network device.
+ * Let the name as short as possible, as it will be part of
+ * memzone name.
+ */
+ char name[RTE_KDP_NAMESIZE];
+ uint32_t core_id; /* Core ID to bind kernel thread on */
+ uint16_t group_id;
+ unsigned mbuf_size;
+
+ uint8_t force_bind; /* Flag to bind kernel thread */
+};
+
+struct rte_kdp *rte_kdp_init(uint16_t port_id);
+
+int rte_kdp_start(struct rte_kdp *kdp, struct rte_mempool *pktmbuf_pool,
+ const struct rte_kdp_conf *conf);
+
+unsigned rte_kdp_rx_burst(struct rte_kdp *kdp,
+ struct rte_mbuf **mbufs, unsigned num);
+
+unsigned rte_kdp_tx_burst(struct rte_kdp *kdp,
+ struct rte_mbuf **mbufs, unsigned num);
+
+int rte_kdp_release(struct rte_kdp *kdp);
+
+void rte_kdp_close(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _RTE_KDP_H_ */
diff --git a/drivers/net/kdp/rte_kdp_fifo.h b/drivers/net/kdp/rte_kdp_fifo.h
new file mode 100644
index 0000000..1a7e063
--- /dev/null
+++ b/drivers/net/kdp/rte_kdp_fifo.h
@@ -0,0 +1,91 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/**
+ * Initializes the kdp fifo structure
+ */
+static void
+kdp_fifo_init(struct rte_kdp_fifo *fifo, unsigned size)
+{
+ /* Ensure size is power of 2 */
+ if (size & (size - 1))
+ rte_panic("KDP fifo size must be power of 2\n");
+
+ fifo->write = 0;
+ fifo->read = 0;
+ fifo->len = size;
+ fifo->elem_size = sizeof(void *);
+}
+
+/**
+ * Adds num elements into the fifo. Return the number actually written
+ */
+static inline unsigned
+kdp_fifo_put(struct rte_kdp_fifo *fifo, void **data, unsigned num)
+{
+ unsigned i = 0;
+ unsigned fifo_write = fifo->write;
+ unsigned fifo_read = fifo->read;
+ unsigned new_write = fifo_write;
+
+ for (i = 0; i < num; i++) {
+ new_write = (new_write + 1) & (fifo->len - 1);
+
+ if (new_write == fifo_read)
+ break;
+ fifo->buffer[fifo_write] = data[i];
+ fifo_write = new_write;
+ }
+ fifo->write = fifo_write;
+ return i;
+}
+
+/**
+ * Get up to num elements from the fifo. Return the number actully read
+ */
+static inline unsigned
+kdp_fifo_get(struct rte_kdp_fifo *fifo, void **data, unsigned num)
+{
+ unsigned i = 0;
+ unsigned new_read = fifo->read;
+ unsigned fifo_write = fifo->write;
+ for (i = 0; i < num; i++) {
+ if (new_read == fifo_write)
+ break;
+
+ data[i] = fifo->buffer[new_read];
+ new_read = (new_read + 1) & (fifo->len - 1);
+ }
+ fifo->read = new_read;
+ return i;
+}
diff --git a/drivers/net/kdp/rte_pmd_kdp_version.map b/drivers/net/kdp/rte_pmd_kdp_version.map
new file mode 100644
index 0000000..0812bb1
--- /dev/null
+++ b/drivers/net/kdp/rte_pmd_kdp_version.map
@@ -0,0 +1,4 @@
+DPDK_2.3 {
+
+ local: *;
+};
diff --git a/lib/librte_eal/common/include/rte_log.h b/lib/librte_eal/common/include/rte_log.h
index 2e47e7f..5a0048b 100644
--- a/lib/librte_eal/common/include/rte_log.h
+++ b/lib/librte_eal/common/include/rte_log.h
@@ -1,7 +1,7 @@
/*-
* BSD LICENSE
*
- * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
+ * Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
@@ -79,6 +79,7 @@ extern struct rte_logs rte_logs;
#define RTE_LOGTYPE_PIPELINE 0x00008000 /**< Log related to pipeline. */
#define RTE_LOGTYPE_MBUF 0x00010000 /**< Log related to mbuf. */
#define RTE_LOGTYPE_CRYPTODEV 0x00020000 /**< Log related to cryptodev. */
+#define RTE_LOGTYPE_KDP 0x00080000 /**< Log related to KDP. */
/* these log types can be used in an application */
#define RTE_LOGTYPE_USER1 0x01000000 /**< User-defined log type 1. */
diff --git a/mk/rte.app.mk b/mk/rte.app.mk
index 8ecab41..eb18972 100644
--- a/mk/rte.app.mk
+++ b/mk/rte.app.mk
@@ -1,6 +1,6 @@
# BSD LICENSE
#
-# Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
+# Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
# Copyright(c) 2014-2015 6WIND S.A.
# All rights reserved.
#
@@ -154,6 +154,7 @@ _LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_PCAP) += -lrte_pmd_pcap
_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_AF_PACKET) += -lrte_pmd_af_packet
_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_NULL) += -lrte_pmd_null
_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_QAT) += -lrte_pmd_qat
+_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_KDP) += -lrte_pmd_kdp
# AESNI MULTI BUFFER is dependent on the IPSec_MB library
_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_AESNI_MB) += -lrte_pmd_aesni_mb
--
2.5.0
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux
2016-01-22 16:00 [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 1/2] kdp: add kernel data path kernel module Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 2/2] kdp: add virtual PMD for kernel slow data path communication Ferruh Yigit
@ 2016-01-22 16:31 ` Thomas Monjalon
2016-01-22 16:56 ` Ferruh Yigit
2 siblings, 1 reply; 7+ messages in thread
From: Thomas Monjalon @ 2016-01-22 16:31 UTC (permalink / raw)
To: Ferruh Yigit; +Cc: dev
Hi Ferruh,
Not commenting the implementation, just the method.
2016-01-22 16:00, Ferruh Yigit:
> This is slow data path communication implementation based on existing KNI.
> Difference is: librte_kni converted into a PMD, kdp kernel module is almost
> same except all control path functionality removed and some simplification done.
Is there a chance to submit such kernel module on LKML instead of DPDK?
We should avoid maintaining some out-of-tree modules.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux
2016-01-22 16:31 ` [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Thomas Monjalon
@ 2016-01-22 16:56 ` Ferruh Yigit
2016-01-22 17:15 ` Thomas Monjalon
0 siblings, 1 reply; 7+ messages in thread
From: Ferruh Yigit @ 2016-01-22 16:56 UTC (permalink / raw)
To: Thomas Monjalon; +Cc: dev
On Fri, Jan 22, 2016 at 05:31:45PM +0100, Thomas Monjalon wrote:
> Hi Ferruh,
>
> Not commenting the implementation, just the method.
>
> 2016-01-22 16:00, Ferruh Yigit:
> > This is slow data path communication implementation based on existing KNI.
> > Difference is: librte_kni converted into a PMD, kdp kernel module is almost
> > same except all control path functionality removed and some simplification done.
>
> Is there a chance to submit such kernel module on LKML instead of DPDK?
> We should avoid maintaining some out-of-tree modules.
The ones I have sent are not generic enough to be in Linux tree.
We already maintain kni kernel module, these patches are part of effort to make
kni more maintainable, by separation of concerns, removing network drivers from it,
and simplifying some of code.
For this patch set, tun/tap interface can be alternative, and it looks like it
removes out-of-tree kernel module requirement, unless people want current
FIFO implementation because of better performance.
For control path, unfortunately I am not aware of any solution without out-of-tree
kernel module support.
Thanks,
ferruh
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux
2016-01-22 16:56 ` Ferruh Yigit
@ 2016-01-22 17:15 ` Thomas Monjalon
2016-01-23 16:14 ` Bruce Richardson
0 siblings, 1 reply; 7+ messages in thread
From: Thomas Monjalon @ 2016-01-22 17:15 UTC (permalink / raw)
To: Ferruh Yigit; +Cc: dev
2016-01-22 16:56, Ferruh Yigit:
> On Fri, Jan 22, 2016 at 05:31:45PM +0100, Thomas Monjalon wrote:
> > Hi Ferruh,
> >
> > Not commenting the implementation, just the method.
> >
> > 2016-01-22 16:00, Ferruh Yigit:
> > > This is slow data path communication implementation based on existing KNI.
> > > Difference is: librte_kni converted into a PMD, kdp kernel module is almost
> > > same except all control path functionality removed and some simplification done.
> >
> > Is there a chance to submit such kernel module on LKML instead of DPDK?
> > We should avoid maintaining some out-of-tree modules.
>
> The ones I have sent are not generic enough to be in Linux tree.
I've not read the details.
What is missing to be generic?
> We already maintain kni kernel module,
Yes it is painful and not accepted in some Linux distros.
> these patches are part of effort to make
> kni more maintainable, by separation of concerns, removing network drivers from it,
> and simplifying some of code.
Your patch is not removing KNI unfortunately ;)
> For this patch set, tun/tap interface can be alternative, and it looks like it
> removes out-of-tree kernel module requirement, unless people want current
> FIFO implementation because of better performance.
>
> For control path, unfortunately I am not aware of any solution without out-of-tree
> kernel module support.
>
> Thanks,
> ferruh
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux
2016-01-22 17:15 ` Thomas Monjalon
@ 2016-01-23 16:14 ` Bruce Richardson
0 siblings, 0 replies; 7+ messages in thread
From: Bruce Richardson @ 2016-01-23 16:14 UTC (permalink / raw)
To: Thomas Monjalon; +Cc: dev
On Fri, Jan 22, 2016 at 06:15:40PM +0100, Thomas Monjalon wrote:
> 2016-01-22 16:56, Ferruh Yigit:
> > On Fri, Jan 22, 2016 at 05:31:45PM +0100, Thomas Monjalon wrote:
> > > Hi Ferruh,
> > >
> > > Not commenting the implementation, just the method.
> > >
> > > 2016-01-22 16:00, Ferruh Yigit:
> > > > This is slow data path communication implementation based on existing KNI.
> > > > Difference is: librte_kni converted into a PMD, kdp kernel module is almost
> > > > same except all control path functionality removed and some simplification done.
> > >
> > > Is there a chance to submit such kernel module on LKML instead of DPDK?
> > > We should avoid maintaining some out-of-tree modules.
> >
> > The ones I have sent are not generic enough to be in Linux tree.
>
> I've not read the details.
> What is missing to be generic?
>
> > We already maintain kni kernel module,
>
> Yes it is painful and not accepted in some Linux distros.
>
> > these patches are part of effort to make
> > kni more maintainable, by separation of concerns, removing network drivers from it,
> > and simplifying some of code.
>
> Your patch is not removing KNI unfortunately ;)
>
One step at a time. :-) I think we all want to get there.
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2016-01-23 16:14 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-01-22 16:00 [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 1/2] kdp: add kernel data path kernel module Ferruh Yigit
2016-01-22 16:00 ` [dpdk-dev] [RFC 2/2] kdp: add virtual PMD for kernel slow data path communication Ferruh Yigit
2016-01-22 16:31 ` [dpdk-dev] [RFC 0/2] slow data path communication between DPDK port and Linux Thomas Monjalon
2016-01-22 16:56 ` Ferruh Yigit
2016-01-22 17:15 ` Thomas Monjalon
2016-01-23 16:14 ` Bruce Richardson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).