Date: Sun, 14 Sep 2014 08:42:54 -0400 (EDT) From: Rick Macklem <rmacklem@uoguelph.ca> To: Hans Petter Selasky <hps@selasky.org> Cc: svn-src-head@freebsd.org, Adrian Chadd <adrian@freebsd.org> Subject: Re: svn commit: r271504 - in head/sys: dev/oce dev/vmware/vmxnet3 dev/xen/netfront net netinet ofed/drivers/net/mlx4 Message-ID: <25891118.35942355.1410698574131.JavaMail.root@uoguelph.ca>
next in thread | raw e-mail | index | archive | help
Hans Petter Selasky wrote: [stuff snipped for brevity] > > Btw, your patch calls sbsndptr() in tcp_output(), which advances > > sb_sndptroff and sb_sndptr by the length. > > Then it loops around and reduces the length for the case where > > there are too many mbufs in the chain. > > > > Right, though this patch would need to understand segment lengths too > and not only count them. Yep. I didn't mean that you would want to use the patch, I was just suggesting that you might want to consider doing something like sbsnfmbuf() so that sb_sndptroff and sb_sndptr aren't being advanced in your patch. All I did to make sbsndmbuf() was clone sbsndptr(), then take out the code that updated sb_sndptroff and sb_sndptr plus add a little bit that I found useful for my patch. As an aside, although allowing specification of a limit for segment size sounds like a good plan, I am not aware of hardware that can't handle a large segment? Good luck with it, rick
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?25891118.35942355.1410698574131.JavaMail.root>