From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [patch 3/5] KVM: hypercall batching Date: Tue, 19 Feb 2008 10:30:00 +0200 Message-ID: <47BA9388.5090908@qumranet.com> References: <20080216220924.733723618@redhat.com> > <20080216221220.924823582@redhat.com>> <1203273628.24928.2.camel@diesel> <47B93C70.1010303@qumranet.com> <1203361276.3428.6.camel@basalt> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Cc: kvm-devel@lists.sourceforge.net, Marcelo Tosatti To: Hollis Blanchard Return-path: In-Reply-To: <1203361276.3428.6.camel@basalt> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: kvm-devel-bounces@lists.sourceforge.net Errors-To: kvm-devel-bounces@lists.sourceforge.net List-Id: kvm.vger.kernel.org Hollis Blanchard wrote: >> Kernel data is physically contiguous (true for per-cpu data as well?), >> so no there's issue here. >> > > So this is an addition to the ABI, that the data must be physically > contiguous. That's a pretty subtle implicit requirement, and it's easy > to resolve the issue by requiring page-aligned data in the guest in the > first place. > > Please don't forget that there are other OSes that could use this > interface as well... > I think it's fairly straightforward that a (phys_addr, length) API implies physically contiguous memory. If the guest doesn't have physically contiguous memory, it can page align itself. But you do raise an important issue, that our ABIs (host/user and host/guest) are seriously underdocumented. -- error compiling committee.c: too many arguments to function ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/