Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 18 Jul 2007 13:36:11 +0900
From:      Pyun YongHyeon <pyunyh@gmail.com>
To:        "Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net>
Cc:        freebsd-net@freebsd.org
Subject:   Re: soekris/sis tx checksum problems
Message-ID:  <20070718043611.GD37935@cdnetworks.co.kr>
In-Reply-To: <20070716201218.U31116@maildrop.int.zabbadoz.net>
References:  <4678896b.1cef600a.1a79.7312@mx.google.com> <20070620100441.E98813@maildrop.int.zabbadoz.net> <4ab61a80706200336l49f16764t1d95c61f0dd323e5@mail.gmail.com> <20070716181837.Y31116@maildrop.int.zabbadoz.net> <4ab61a80707161233m11e49f2brac162a882f4ebbd7@mail.gmail.com> <20070716201218.U31116@maildrop.int.zabbadoz.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jul 16, 2007 at 08:14:14PM +0000, Bjoern A. Zeeb wrote:
 > On Mon, 16 Jul 2007, David Cornejo wrote:
 > 
 > >I eventually reinstalled the OS from a recent snapshot then updated it
 > >to the latest CURRENT and it works fine now :-(
 > 
 > Hmm I am using CURRENT from today and I culd see the issue.
 > 
 > I set sysctl net.inet.udp.checksum=0 to get things working temporary.
 > 
 > In addition I am backing out a change from May and shall see if that
 > helps.
 > 

I've looked over sis_encap() and it seems that there is potential
bug for descriptor wrap-around/shortage in bus_dmamap_load loop.
After checking for necessity of m_defrag(9) it blindly assumes
it has enough available Tx descriptors to send the packet.
What if it sill have more fragments than currently available Tx
desciptors after calling m_defrag()?

Unfortunately bus_dma(9) changes for sis(4) requires real hardware
and I don't have it... If I had it I would have fixed it to run
it on sparc64.

 > Did you have/do you have any private patches in your tree?
 > 
-- 
Regards,
Pyun YongHyeon



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070718043611.GD37935>