DPDK patches and discussions
 help / color / mirror / Atom feed
From: David Hunt <david.hunt@intel.com>
To: dev@dpdk.org
Cc: ben.boren@intel.com
Subject: [dpdk-dev] [PATCH 00/11] add armv8 architecture support
Date: Fri, 23 Oct 2015 15:17:02 +0100	[thread overview]
Message-ID: <1445609833-17649-1-git-send-email-david.hunt@intel.com> (raw)

This patch provides support for the ARMv8 architecture. We hope that this will
encourage the ARM community to contribute PMDs for their SoCs to DPDK.

For now, we've added Intel engineers to the MAINTAINERS file. We would like to
encourage the ARM community to take over maintenance of this area in future,
and to further improve it.

ACL and LPM libraries have been disabled in the config due to their reliance
on SSE. 

This patch was tested on an Applied Micro X-Gene Mustang board (pcap only)

This patch was also tested with an Intel 82599 NIC on a Gigabyte MP30-AR0
passing traffic between the two NIC ports using testpmd.

Notes on DPDK configuration:

  Did not use igb_uio or kni so the following optionse were disabled in 
  config/common_linuxapp

  CONFIG_RTE_EAL_IGB_UIO=n
  CONFIG_RTE_LIBRTE_KNI=n
  CONFIG_RTE_KNI_KMOD=n

  make config T=arm64-native-linuxapp-gcc
  make install T=arm64-native-linuxapp-gcc
  
Notes on arm64 kernel configuration:

  Using Ubuntu 14.04 LTS with a 4.3.0-rc6 kernel (with modified PCI drivers),
  and uio_pci_generic.
  ARM64 kernels do not seem to have functional resource mapping of PCI memory
  (PCI_MMAP), so the pci driver needs to be patched to enable this. The
  symptom of this is when /sys/bus/pci/devices/0000:0X:00.Y directory is
  missing the resource0...N files for mmapping the device memory. Earlier
  kernels (3.13.x) had these files present, but mmap'ping resulted in a
  "Bus Error" when the NIC memory was accessed.
  However, during limited testing with a modified 4.3.0-rc6 kernel, we were
  able to mmap the NIC memory, and pass traffic between the two ports on a
  82599 NIC connected via fibre cable. 
  We have no plans to upstream a kernel patch for this and hope that
  someone more familiar with the arm architecture can create a proper patch
  and enable this functionality.

Benjamin Boren (11):
  lib: add armv8 rte_atomic.h
  lib: add armv8 rte_byteorder.h
  lib: add armv8 rte_cpuflags.h
  lib: add armv8 rte_cycles.h
  lib: add armv8 rte_memcpy.h
  lib: add armv8 rte_prefetch.h
  lib: add armv8 rte_rwlock.h
  lib: add armv8 rte_spinlock.h
  lib: add armv8 rte_vect.h
  mk: add makefile and config changes for armv8 architecture
  app: add armv8 cpuflags check to test_cpuflags application

 MAINTAINERS                                        |   6 +
 app/test/test_cpuflags.c                           |  17 ++
 config/defconfig_arm64-native-linuxapp-gcc         |  55 +++++
 .../common/include/arch/arm64/rte_atomic.h         | 269 +++++++++++++++++++++
 .../common/include/arch/arm64/rte_byteorder.h      | 157 ++++++++++++
 .../common/include/arch/arm64/rte_cpuflags.h       | 137 +++++++++++
 .../common/include/arch/arm64/rte_cycles.h         |  77 ++++++
 .../common/include/arch/arm64/rte_memcpy.h         | 266 ++++++++++++++++++++
 .../common/include/arch/arm64/rte_prefetch.h       |  61 +++++
 .../common/include/arch/arm64/rte_rwlock.h         |  70 ++++++
 .../common/include/arch/arm64/rte_spinlock.h       | 114 +++++++++
 .../common/include/arch/arm64/rte_vect.h           | 102 ++++++++
 mk/arch/arm64/rte.vars.mk                          |  58 +++++
 mk/machine/armv8-a/rte.vars.mk                     |  57 +++++
 mk/rte.cpuflags.mk                                 |   9 +
 15 files changed, 1455 insertions(+)
 create mode 100644 config/defconfig_arm64-native-linuxapp-gcc
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_atomic.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_byteorder.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_cpuflags.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_cycles.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_memcpy.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_prefetch.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_rwlock.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_spinlock.h
 create mode 100644 lib/librte_eal/common/include/arch/arm64/rte_vect.h
 create mode 100644 mk/arch/arm64/rte.vars.mk
 create mode 100644 mk/machine/armv8-a/rte.vars.mk

-- 
2.1.4

             reply	other threads:[~2015-10-23 14:17 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-10-23 14:17 David Hunt [this message]
2015-10-23 14:17 ` [dpdk-dev] [PATCH 01/11] lib: add armv8 rte_atomic.h David Hunt
2015-10-23 14:17 ` [dpdk-dev] [PATCH 02/11] lib: add armv8 rte_byteorder.h David Hunt
2015-10-23 14:17 ` [dpdk-dev] [PATCH 03/11] lib: add armv8 rte_cpuflags.h David Hunt
2015-10-26 16:23   ` Jan Viktorin
2015-10-23 14:17 ` [dpdk-dev] [PATCH 04/11] lib: add armv8 rte_cycles.h David Hunt
2015-10-23 14:17 ` [dpdk-dev] [PATCH 05/11] lib: add armv8 rte_memcpy.h David Hunt
2015-10-26 16:22   ` Jan Viktorin
2015-10-23 14:17 ` [dpdk-dev] [PATCH 06/11] lib: add armv8 rte_prefetch.h David Hunt
2015-10-23 14:17 ` [dpdk-dev] [PATCH 07/11] lib: add armv8 rte_rwlock.h David Hunt
2015-10-23 14:17 ` [dpdk-dev] [PATCH 08/11] lib: add armv8 rte_spinlock.h David Hunt
2015-10-23 14:17 ` [dpdk-dev] [PATCH 09/11] lib: add armv8 rte_vect.h David Hunt
2015-10-26 16:23   ` Jan Viktorin
2015-10-23 14:17 ` [dpdk-dev] [PATCH 10/11] mk: add makefile and config changes for armv8 architecture David Hunt
2015-10-26 16:22   ` Jan Viktorin
2015-10-27 18:24     ` Jan Viktorin
2015-10-26 16:22   ` Jan Viktorin
2015-10-23 14:17 ` [dpdk-dev] [PATCH 11/11] app: add armv8 cpuflags check to test_cpuflags application David Hunt
2015-10-26 16:21 ` [dpdk-dev] [PATCH 00/11] add armv8 architecture support Jan Viktorin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1445609833-17649-1-git-send-email-david.hunt@intel.com \
    --to=david.hunt@intel.com \
    --cc=ben.boren@intel.com \
    --cc=dev@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).