From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alex Williamson Subject: Re: [RFC PATCH 3/6] kvm: Merge id_to_index into memslots Date: Wed, 05 Dec 2012 15:58:22 -0700 Message-ID: <1354748302.3224.58.camel@bling.home> References: <20121203231912.3661.57179.stgit@bling.home> <20121203233924.3661.81373.stgit@bling.home> <20121205212200.GB20260@amt.cnet> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: kvm@vger.kernel.org, gleb@redhat.com, linux-kernel@vger.kernel.org To: Marcelo Tosatti Return-path: In-Reply-To: <20121205212200.GB20260@amt.cnet> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On Wed, 2012-12-05 at 19:22 -0200, Marcelo Tosatti wrote: > On Mon, Dec 03, 2012 at 04:39:24PM -0700, Alex Williamson wrote: > > This allows us to resize this structure and therefore the number of > > memslots as part of the RCU update. > > Why is this necessary? "struct memslots" is updated, message above > conflicts with that. > > If there is a reason, "id_to_index" becomes confusing. > I can't have two indeterminate sized arrays in the same structure, so by moving id_to_index into struct kvm_memory_slot I get back to one array. I'm playing with whether this is worthwhile. If we reduce id_to_index to a short then we can leave it with a static size and still get a memory savings for the typical case. Thanks, Alex