From mboxrd@z Thu Jan 1 00:00:00 1970 From: Xiao Guangrong Subject: Re: [PATCH v3 07/15] KVM: MMU: introduce nulls desc Date: Mon, 25 Nov 2013 14:29:03 +0800 Message-ID: <5292EE2F.5090305@linux.vnet.ibm.com> References: <1382534973-13197-1-git-send-email-xiaoguangrong@linux.vnet.ibm.com> <1382534973-13197-8-git-send-email-xiaoguangrong@linux.vnet.ibm.com> <20131122191429.GA13308@amt.cnet> <65EE805B-B5DB-4BD0-A057-E5FF78D96D67@linux.vnet.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Gleb Natapov , avi.kivity@gmail.com, "pbonzini@redhat.com Bonzini" , linux-kernel@vger.kernel.org, kvm@vger.kernel.org, Eric Dumazet , Peter Zijlstra To: Xiao Guangrong , Marcelo Tosatti Return-path: In-Reply-To: <65EE805B-B5DB-4BD0-A057-E5FF78D96D67@linux.vnet.ibm.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 11/25/2013 02:11 PM, Xiao Guangrong wrote: >=20 > On Nov 23, 2013, at 3:14 AM, Marcelo Tosatti wr= ote: >=20 >> On Wed, Oct 23, 2013 at 09:29:25PM +0800, Xiao Guangrong wrote: >>> It likes nulls list and we use the pte-list as the nulls which can = help us to >>> detect whether the "desc" is moved to anther rmap then we can re-wa= lk the rmap >>> if that happened >>> >>> kvm->slots_lock is held when we do lockless walking that prevents r= map >>> is reused (free rmap need to hold that lock) so that we can not see= the same >>> nulls used on different rmaps >>> >>> Signed-off-by: Xiao Guangrong >> >> How about simplified lockless walk on the slot while rmapp entry >> contains a single spte? (which should be the case with two-dimension= al >> paging). >> >> That is, grab the lock when finding a rmap with more than one spte i= n >> it (and then keep it locked until the end). >=20 > Hmm=EF=BF=BD that isn't straightforward and more complex than the app= roach > in this patchset. Also it can drop the improvement for shadow mmu tha= t > gets great improvement by this patchset. >=20 >> >> For example, nothing prevents lockless walker to move into some >> parent_ptes chain, right? >=20 > No. >=20 > The nulls can help us to detect this case, for parent_ptes, the nulls= points > to "shadow page" but for rmaps, the nulls points to slot.arch.rmap. T= here > is no chance that the =EF=BF=BDrmap" is used as shadow page when slot= -lock is held. >=20 >> >> Also, there is no guarantee of termination (as long as sptes are >> deleted with the correct timing). BTW, can't see any guarantee of >> termination for rculist nulls either (a writer can race with a lockl= ess >> reader indefinately, restarting the lockless walk every time). >=20 > Hmm, that can be avoided by checking dirty-bitmap before rewalk, > that means, if the dirty-bitmap has been set during lockless write-pr= otection, > it=EF=BF=BDs unnecessary to write-protect its sptes. Your idea? This idea is based on the fact that the number of rmap is limited by RMAP_RECYCLE_THRESHOLD. So, in the case of adding new spte into rmap, we can break the rewalk at once, in the case of deleting, we can only rewalk RMAP_RECYCLE_THRESHOLD times.