From patchwork Mon Aug 2 08:15:08 2010 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Xiao Guangrong X-Patchwork-Id: 116452 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by demeter.kernel.org (8.14.4/8.14.3) with ESMTP id o728Jg6T022884 for ; Mon, 2 Aug 2010 08:19:42 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753335Ab0HBITY (ORCPT ); Mon, 2 Aug 2010 04:19:24 -0400 Received: from cn.fujitsu.com ([222.73.24.84]:62948 "EHLO song.cn.fujitsu.com" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1751464Ab0HBITX (ORCPT ); Mon, 2 Aug 2010 04:19:23 -0400 Received: from tang.cn.fujitsu.com (tang.cn.fujitsu.com [10.167.250.3]) by song.cn.fujitsu.com (Postfix) with ESMTP id B454E170130; Mon, 2 Aug 2010 16:19:20 +0800 (CST) Received: from fnst.cn.fujitsu.com (tang.cn.fujitsu.com [127.0.0.1]) by tang.cn.fujitsu.com (8.14.3/8.13.1) with ESMTP id o728GJbV019270; Mon, 2 Aug 2010 16:16:19 +0800 Received: from [10.167.141.99] (unknown [10.167.141.99]) by fnst.cn.fujitsu.com (Postfix) with ESMTPA id 4622B14C00D; Mon, 2 Aug 2010 16:20:04 +0800 (CST) Message-ID: <4C567E8C.3090305@cn.fujitsu.com> Date: Mon, 02 Aug 2010 16:15:08 +0800 From: Xiao Guangrong User-Agent: Thunderbird 2.0.0.24 (Windows/20100228) MIME-Version: 1.0 To: Avi Kivity CC: Marcelo Tosatti , LKML , KVM list Subject: [PATCH v2 3/3] KVM: MMU: mark page dirty only when page is really written References: <4C4E5313.1080308@cn.fujitsu.com> <4C4E536E.7030109@cn.fujitsu.com> <4C4E53F7.7050108@cn.fujitsu.com> In-Reply-To: <4C4E53F7.7050108@cn.fujitsu.com> Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org X-Greylist: IP, sender and recipient auto-whitelisted, not delayed by milter-greylist-4.2.3 (demeter.kernel.org [140.211.167.41]); Mon, 02 Aug 2010 08:19:42 +0000 (UTC) diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index e18834c..9c69725 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -311,24 +311,42 @@ static bool spte_has_volatile_bits(u64 spte) if (!is_shadow_present_pte(spte)) return false; - if (spte & shadow_accessed_mask) + if ((spte & shadow_accessed_mask) && + (!is_writable_pte(spte) || (spte & shadow_dirty_mask))) return false; return true; } +static bool spte_is_bit_cleared(u64 old_spte, u64 new_spte, u64 bit_mask) +{ + return (old_spte & bit_mask) && !(new_spte & bit_mask); +} + static void update_spte(u64 *sptep, u64 new_spte) { - u64 old_spte; + u64 mask, old_spte = *sptep; + + WARN_ON(!is_rmap_spte(new_spte)); - if (!shadow_accessed_mask || (new_spte & shadow_accessed_mask) || - !is_rmap_spte(*sptep)) + new_spte |= old_spte & shadow_dirty_mask; + + mask = shadow_accessed_mask; + if (is_writable_pte(old_spte)) + mask |= shadow_dirty_mask; + + if (!spte_has_volatile_bits(old_spte) || (new_spte & mask) == mask) __set_spte(sptep, new_spte); - else { + else old_spte = __xchg_spte(sptep, new_spte); - if (old_spte & shadow_accessed_mask) - kvm_set_pfn_accessed(spte_to_pfn(old_spte)); - } + + if (!shadow_accessed_mask) + return; + + if (spte_is_bit_cleared(old_spte, new_spte, shadow_accessed_mask)) + kvm_set_pfn_accessed(spte_to_pfn(old_spte)); + if (spte_is_bit_cleared(old_spte, new_spte, shadow_dirty_mask)) + kvm_set_pfn_dirty(spte_to_pfn(old_spte)); } static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache, @@ -708,7 +726,7 @@ static void set_spte_track_bits(u64 *sptep, u64 new_spte) pfn = spte_to_pfn(old_spte); if (!shadow_accessed_mask || old_spte & shadow_accessed_mask) kvm_set_pfn_accessed(pfn); - if (is_writable_pte(old_spte)) + if (!shadow_dirty_mask || (old_spte & shadow_dirty_mask)) kvm_set_pfn_dirty(pfn); } @@ -763,13 +781,6 @@ static int rmap_write_protect(struct kvm *kvm, u64 gfn) } spte = rmap_next(kvm, rmapp, spte); } - if (write_protected) { - pfn_t pfn; - - spte = rmap_next(kvm, rmapp, NULL); - pfn = spte_to_pfn(*spte); - kvm_set_pfn_dirty(pfn); - } /* check for huge page mappings */ for (i = PT_DIRECTORY_LEVEL; @@ -1942,7 +1953,7 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, * whether the guest actually used the pte (in order to detect * demand paging). */ - spte = shadow_base_present_pte | shadow_dirty_mask; + spte = shadow_base_present_pte; if (!speculative) spte |= shadow_accessed_mask; if (!dirty) @@ -2003,8 +2014,6 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, mark_page_dirty(vcpu->kvm, gfn); set_pte: - if (is_writable_pte(*sptep) && !is_writable_pte(spte)) - kvm_set_pfn_dirty(pfn); update_spte(sptep, spte); done: return ret;