Date: Tue, 08 Mar 2005 22:08:54 -0700 From: Scott Long <scottl@samsco.org> To: Mike Silbersack <silby@silby.com> Cc: cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/pci if_dc.c Message-ID: <422E84E6.1030508@samsco.org> In-Reply-To: <20050308194137.J811@odysseus.silby.com> References: <200503082325.j28NPkY5032559@repoman.freebsd.org> <20050308194137.J811@odysseus.silby.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Mike Silbersack wrote: > > On Tue, 8 Mar 2005, Scott Long wrote: > >> The DC driver asks for an alignment of PAGE_SIZE for data buffers, >> but also >> asks that each buffer be (2048 * 256) bytes long. I suspect that >> alignment >> isn't a real requirement since busdma only recently started honoring >> it. The >> size is also bogus. Fix both of these and stop busdma from trying to >> exhaust the system memory pool with bounce pages. > > > The requirement as it was originally stated is that they may need to be > longword aligned for certain cards. We handle that in dc_encap already, > but I hadn't considered the possible problem with using bounce pages. > > I'll take a look and see if we can determine the alignment requirement > before we call bus_dma_tag_create. > > Are there any busdma statistics sysctls that would let us notice that > it's being asked to do things like this on a running system? If not, > could you add some? > > Mike "Silby" Silbersack Yes, the hw.busdma tree has quite a bit of stats regarding bounce page use. Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?422E84E6.1030508>