DPDK patches and discussions
 help / color / mirror / Atom feed
From: Harry van Haaren <harry.van.haaren@intel.com>
To: john.mcnamara@intel.com
Cc: dev@dpdk.org
Subject: [dpdk-dev] [PATCH v3] docs: add statistics read frequency to fm10k guide
Date: Tue,  8 Mar 2016 17:16:35 +0000	[thread overview]
Message-ID: <1457457395-28853-1-git-send-email-harry.van.haaren@intel.com> (raw)
In-Reply-To: <1455031194-27921-1-git-send-email-harry.van.haaren@intel.com>

This patch documents that the statistics of fm10k based NICs must be
read regularly in order to avoid an undetected 32 bit integer-overflow.

Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: John McNamara <john.mcnamara@intel.com>
---

v3:
-Removed paragraph from "Known Issues" based on feedback

v2:
-Moved notes to Known Issues section of Release Notes.

 doc/guides/nics/fm10k.rst | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/doc/guides/nics/fm10k.rst b/doc/guides/nics/fm10k.rst
index 4206b7f..63b495d 100644
--- a/doc/guides/nics/fm10k.rst
+++ b/doc/guides/nics/fm10k.rst
@@ -65,3 +65,19 @@ The FM10000 family of NICS support a maximum of a 15K jumbo frame. The value
 is fixed and cannot be changed. So, even when the ``rxmode.max_rx_pkt_len``
 member of ``struct rte_eth_conf`` is set to a value lower than 15364, frames
 up to 15364 bytes can still reach the host interface.
+
+Statistic Polling Frequency
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The FM10000 NICs expose a set of statistics via the PCI BARs. These statistics
+are read from the hardware registers when ``rte_eth_stats_get()`` or
+``rte_eth_xstats_get()`` is called. The packet counting registers are 32 bits
+while the byte counting registers are 48 bits. As a result, the statistics must
+be polled regularly in order to ensure the consistency of the returned reads.
+
+Given the PCIe Gen3 x8, about 50Gbps of traffic can occur. With 64 byte packets
+this gives almost 100 million packets/second, causing 32 bit integer overflow
+after approx 40 seconds. To ensure these overflows are detected and accounted
+for in the statistics, it is necessary to read statistic regularly. It is
+suggested to read stats every 20 seconds, which will ensure the statistics
+are accurate.
-- 
2.5.0

  parent reply	other threads:[~2016-03-08 17:16 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-08 15:42 [dpdk-dev] [PATCH] " Harry van Haaren
2016-02-08 16:00 ` Thomas Monjalon
2016-02-08 16:06   ` Van Haaren, Harry
2016-02-09 15:00   ` Mcnamara, John
2016-02-09 15:19 ` [dpdk-dev] [PATCH v2] " Harry van Haaren
2016-02-22 17:13   ` Mcnamara, John
2016-02-22 17:18   ` Mcnamara, John
2016-02-22 17:40     ` Thomas Monjalon
2016-03-06 22:35   ` Thomas Monjalon
2016-03-08 17:16   ` Harry van Haaren [this message]
2016-03-09 17:39     ` [dpdk-dev] [PATCH v3] " Thomas Monjalon

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=1457457395-28853-1-git-send-email-harry.van.haaren@intel.com \
    --to=harry.van.haaren@intel.com \
    --cc=dev@dpdk.org \
    --cc=john.mcnamara@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).