DPDK patches and discussions
 help / color / mirror / Atom feed
From: Anatoly Burakov <anatoly.burakov@intel.com>
To: dev@dpdk.org
Cc: john.mcnamara@intel.com, bruce.richardson@intel.com,
	pablo.de.lara.guarch@intel.com, david.hunt@intel.com,
	mohammad.abdul.awal@intel.com
Subject: [dpdk-dev] [RFC 4/9] usertools/lib: support FreeBSD for platform info
Date: Mon, 25 Jun 2018 16:59:41 +0100	[thread overview]
Message-ID: <798fa82633f347495706e964fc44f6ab273d7baa.1529940601.git.anatoly.burakov@intel.com> (raw)
In-Reply-To: <cover.1529940601.git.anatoly.burakov@intel.com>
In-Reply-To: <cover.1529940601.git.anatoly.burakov@intel.com>

This enables FreeBSD support for PlatformInfo part of the
DPDKConfigLib script, thereby enabling FreeBSD support for the
cpu_layout script.

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
---
 usertools/DPDKConfigLib/PlatformInfo.py | 81 ++++++++++++++++++++++++-
 1 file changed, 78 insertions(+), 3 deletions(-)

diff --git a/usertools/DPDKConfigLib/PlatformInfo.py b/usertools/DPDKConfigLib/PlatformInfo.py
index 734d22026..4ca507a37 100755
--- a/usertools/DPDKConfigLib/PlatformInfo.py
+++ b/usertools/DPDKConfigLib/PlatformInfo.py
@@ -6,6 +6,15 @@
 from .Util import *
 import re
 import glob
+import platform
+import subprocess
+import string
+import xml.etree.ElementTree  # for FreeBSD topology parsing
+try:
+    import queue  # python 3
+except ImportError:
+    import Queue as queue  # python 2
+
 
 __SYSFS_CPU_INFO_PATH = "/sys/devices/system/cpu"
 __SYSFS_IOMMU_CLASS_PATH = "/sys/class/iommu"
@@ -26,6 +35,9 @@
 
 __HT_CPU_FLAGS = ["ht"]
 
+__SYSCTL_TOPOLOGY_CMDLINE = ['sysctl', '-b', 'kern.sched.topology_spec']
+__SYSCTL_MEM_CMDLINE = ['sysctl', 'hw.realmem']
+
 
 def _parse_cpuinfo(pinfo):
     core_info_list = []
@@ -107,6 +119,56 @@ def _iommu_is_enabled():
     pass
 
 
+def _parse_sysctl_cpu_topology(pinfo):
+    output = subprocess.check_output(__SYSCTL_TOPOLOGY_CMDLINE)
+
+    # output from sysctl contains null terminator, remove it
+    raw_xml = output[:-1]
+    tree = xml.etree.ElementTree.fromstring(raw_xml)
+    groups = queue.Queue()
+
+    # put first group onto the queue
+    for e in tree.findall('group'):
+        groups.put(e)
+
+    # per-level list of cores
+    levels = {}
+
+    while not groups.empty():
+        group = groups.get()
+        level = int(group.get('level'))
+        cpus = [int(cpu) for cpu in group.find('cpu').text.split(",")]
+        cur_value = levels.setdefault(level, [])
+
+        # store discovered cpu's
+        cur_value.append(cpus)
+        levels[level] = cur_value
+
+        children = group.find('children')
+        if children is not None:
+            for c in children.findall('group'):
+                groups.put(c)
+
+    # find deepest level
+    max_level = max(levels.keys())
+
+    # for each group in the deepest level, take first CPU and make it physical
+    # core id
+    for cpus in levels[max_level]:
+        pinfo.core_map[pinfo.numa_nodes[0], cpus[0]] = cpus
+
+        # also make note of hyperthreading
+        if len(cpus) > 1:
+            pinfo.hyperthreading_supported = True
+
+
+def _parse_sysctl_ram_size():
+    output = subprocess.check_output(__SYSCTL_MEM_CMDLINE)
+    _, mem_str = kv_split(output, ':')
+    mem_amount = int(mem_str) / 1024  # kilobytes
+    return mem_amount
+
+
 class PlatformInfo:
     def __init__(self):
         self.update()
@@ -125,6 +187,19 @@ def reset(self):
 
     def update(self):
         self.reset()
-        _parse_cpuinfo(self)
-        _parse_meminfo(self)
-        self.hugepage_sizes_enabled = _find_enabled_hugepage_sizes()
+        system = platform.system()
+        if system == 'Linux':
+            _parse_cpuinfo(self)
+            _parse_meminfo(self)
+            self.hugepage_sizes_enabled = _find_enabled_hugepage_sizes()
+        elif system == 'FreeBSD':
+            # DPDK doesn't support NUMA on FreeBSD
+            self.numa_nodes = [0]
+            # find number of threads
+            _parse_sysctl_cpu_topology(self)
+            # find RAM size
+            self.ram_size = _parse_sysctl_ram_size()
+            # DPDK doesn't use hugepages on FreeBSD
+            self.hugepage_sizes_supported = []
+            self.hugepage_sizes_enabled = []
+            self.default_hugepage_size = 0
-- 
2.17.1

  parent reply	other threads:[~2018-06-25 15:59 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-25 15:59 [dpdk-dev] [RFC 0/9] Modularize and enhance DPDK Python scripts Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 1/9] usertools: add DPDK config lib python library Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 2/9] usertools/lib: add platform info library Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 3/9] usertools/cpu_layout: rewrite to use DPDKConfigLib Anatoly Burakov
2018-06-25 15:59 ` Anatoly Burakov [this message]
2018-06-25 15:59 ` [dpdk-dev] [RFC 5/9] usertools/lib: add device information library Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 6/9] usertools/devbind: switch to using DPDKConfigLib Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 7/9] usertools/lib: add hugepage information library Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 8/9] usertools: add hugepage info script Anatoly Burakov
2018-06-25 15:59 ` [dpdk-dev] [RFC 9/9] usertools/lib: add GRUB utility library for hugepage config Anatoly Burakov
2018-06-26  1:09   ` Kevin Wilson
2018-06-26  9:05     ` Burakov, Anatoly
2018-08-14 10:11 ` [dpdk-dev] [RFC 0/9] Modularize and enhance DPDK Python scripts Burakov, Anatoly
2018-08-28  8:16   ` Burakov, Anatoly
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 " Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 1/9] usertools: add DPDK config lib python library Anatoly Burakov
2018-11-16  0:45   ` Stephen Hemminger
2018-11-16 11:49     ` Burakov, Anatoly
2018-11-16 14:09       ` Wiles, Keith
2018-11-16 14:13         ` Richardson, Bruce
2018-11-16 14:37           ` Burakov, Anatoly
2018-11-16 14:55             ` Thomas Monjalon
2018-11-16 15:41               ` Wiles, Keith
2018-11-16 15:43               ` Burakov, Anatoly
2018-11-16 15:58                 ` Thomas Monjalon
2018-11-16 16:10                   ` Bruce Richardson
2018-11-16 16:08                 ` Bruce Richardson
2018-11-16 15:38             ` Wiles, Keith
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 2/9] usertools/lib: add platform info library Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 3/9] usertools/cpu_layout: rewrite to use DPDKConfigLib Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 4/9] usertools/lib: support FreeBSD for platform info Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 5/9] usertools/lib: add device information library Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 6/9] usertools/devbind: switch to using DPDKConfigLib Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 7/9] usertools/lib: add hugepage information library Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 8/9] usertools: add hugepage info script Anatoly Burakov
2018-11-15 15:47 ` [dpdk-dev] [RFC v2 9/9] usertools/lib: add GRUB utility library for hugepage config Anatoly Burakov

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=798fa82633f347495706e964fc44f6ab273d7baa.1529940601.git.anatoly.burakov@intel.com \
    --to=anatoly.burakov@intel.com \
    --cc=bruce.richardson@intel.com \
    --cc=david.hunt@intel.com \
    --cc=dev@dpdk.org \
    --cc=john.mcnamara@intel.com \
    --cc=mohammad.abdul.awal@intel.com \
    --cc=pablo.de.lara.guarch@intel.com \
    /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).