Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 11 Aug 2008 11:17:03 +0900
From:      Pyun YongHyeon <pyunyh@gmail.com>
To:        Christian Weisgerber <naddy@mips.inka.de>
Cc:        freebsd-net@freebsd.org
Subject:   Re: Rx/tx hardware checksumming statistics?
Message-ID:  <20080811021703.GC50045@cdnetworks.co.kr>
In-Reply-To: <g7ktnq$glp$1@lorvorc.mips.inka.de>
References:  <g7ktnq$glp$1@lorvorc.mips.inka.de>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Aug 09, 2008 at 08:12:42PM +0000, Christian Weisgerber wrote:
 > OpenBSD keeps count of the packets that have undergone IPv4 header/
 > TCP/UDP checksumming in hardware.  These statics are available with
 > netstat -s, e.g.:
 > 
 > ip:
 >     ...
 >         492152 input datagrams checksum-processed by hardware
 >         911338 output datagrams checksum-processed by hardware
 >     ...
 > 
 > This comes in quite handy to check whether checksum offloading
 > actually works and which protocols are successfully processed this
 > way.
 > 

I don't think it indicates whether checksum offloading actually
works as OpenBSD blindly set a flag, which was derived from
hardware, to indicate hardware performed the checksum computation.
You can still have a chance that checksum offloading does not work
even if the counter constantly increase over time. In addition,
OpenBSD have a no way to disable checksum offloading feature of
hardware on the fly so you have to rebuild the driver to disable
checksum offloading in case of broken/buggy hardware.

 > On FreeBSD, netstat -s does not provide this information.  Are these
 > statistics available in some other way?  How would I check whether

AFAIK there is no such counters in FreeBSD.

 > packets have actually been checksummed in hardware?
 > 

If checksum offloading didn't work you couldn't get a working
connection at all.

-- 
Regards,
Pyun YongHyeon



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080811021703.GC50045>