Date: Wed, 13 Sep 2000 21:44:25 -0400 (EDT) From: Bosko Milekic <bmilekic@dsuper.net> To: Alfred Perlstein <bright@wintelcom.net> Cc: freebsd-net@FreeBSD.ORG Subject: Re: Clusters larger than PAGE_SIZE and contigmalloc() Message-ID: <Pine.BSF.4.21.0009132132340.9365-100000@jehovah.technokratis.com> In-Reply-To: <20000913171611.E12231@fw.wintelcom.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 13 Sep 2000, Alfred Perlstein wrote: > Well I attempted to use clusters larger than PAGE_SIZE without > contigmalloc not realizing that there was no bus_space_foo for > mbufs. It wasn't fun debugging that. Mike Smith suggested that > I investigate how NetBSD handles this situation, I looked and it > seemed somewhat ok, but at a glance somewhat inneficient. If I recall correctly, you tried allocating your own external buffer. I'm actually wondering whether the kproc that is initialized in the case where MCLBYTES (the constant) is increased by the developer/whoever to something > PAGE_SIZE is still being used, and what the results are if it is. That's the part I'd like to scrap (you see a lot of #if MCLBYTES > PAGE_SIZE junk around the mbuf code). > Well, if you could do the bus space stuff for mbufs that would be > optimal, I'm sure the ethernet driver authors wouldn't have much > trouble migrating over to it. The drivers that typically require these larger buffers are usually the gigabit network adapters that do jumbo frames. I think that some of this hardware actually supports DMAing into several different buffer areas - I wonder if someone more familiar with this (like one of the guys working on the socket zero-copy code, or Bill Paul) could confirm this... > -Alfred Cheers, Bosko Milekic bmilekic@technokratis.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0009132132340.9365-100000>