From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756021AbYLIU2a (ORCPT ); Tue, 9 Dec 2008 15:28:30 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755321AbYLIUXf (ORCPT ); Tue, 9 Dec 2008 15:23:35 -0500 Received: from mx2.redhat.com ([66.187.237.31]:34864 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754769AbYLIUXK (ORCPT ); Tue, 9 Dec 2008 15:23:10 -0500 From: Avi Kivity To: kvm@vger.kernel.org Cc: linux-kernel@vger.kernel.org, Hollis Blanchard Subject: [PATCH 14/44] KVM: ppc: use prefetchable mappings for guest memory Date: Tue, 9 Dec 2008 22:22:29 +0200 Message-Id: <1228854179-23002-15-git-send-email-avi@redhat.com> In-Reply-To: <1228854179-23002-1-git-send-email-avi@redhat.com> References: <1228854179-23002-1-git-send-email-avi@redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: Hollis Blanchard Bare metal Linux on 440 can "overmap" RAM in the kernel linear map, so that it can use large (256MB) mappings even if memory isn't a multiple of 256MB. To prevent the hardware prefetcher from loading from an invalid physical address through that mapping, it's marked Guarded. However, KVM must ensure that all guest mappings are backed by real physical RAM (since a deliberate access through a guarded mapping could still cause a machine check). Accordingly, we don't need to make our mappings guarded, so let's allow prefetching as the designers intended. Curiously this patch didn't affect performance at all on the quick test I tried, but it's clearly the right thing to do anyways and may improve other workloads. Signed-off-by: Hollis Blanchard Signed-off-by: Avi Kivity --- arch/powerpc/kvm/44x_tlb.c | 9 +++++++-- 1 files changed, 7 insertions(+), 2 deletions(-) diff --git a/arch/powerpc/kvm/44x_tlb.c b/arch/powerpc/kvm/44x_tlb.c index 260fa8b..6fadbd6 100644 --- a/arch/powerpc/kvm/44x_tlb.c +++ b/arch/powerpc/kvm/44x_tlb.c @@ -28,6 +28,8 @@ #include "44x_tlb.h" +#define PPC44x_TLB_UATTR_MASK \ + (PPC44x_TLB_U0|PPC44x_TLB_U1|PPC44x_TLB_U2|PPC44x_TLB_U3) #define PPC44x_TLB_USER_PERM_MASK (PPC44x_TLB_UX|PPC44x_TLB_UR|PPC44x_TLB_UW) #define PPC44x_TLB_SUPER_PERM_MASK (PPC44x_TLB_SX|PPC44x_TLB_SR|PPC44x_TLB_SW) @@ -63,8 +65,8 @@ void kvmppc_dump_tlbs(struct kvm_vcpu *vcpu) static u32 kvmppc_44x_tlb_shadow_attrib(u32 attrib, int usermode) { - /* Mask off reserved bits. */ - attrib &= PPC44x_TLB_PERM_MASK|PPC44x_TLB_ATTR_MASK; + /* We only care about the guest's permission and user bits. */ + attrib &= PPC44x_TLB_PERM_MASK|PPC44x_TLB_UATTR_MASK; if (!usermode) { /* Guest is in supervisor mode, so we need to translate guest @@ -76,6 +78,9 @@ static u32 kvmppc_44x_tlb_shadow_attrib(u32 attrib, int usermode) /* Make sure host can always access this memory. */ attrib |= PPC44x_TLB_SX|PPC44x_TLB_SR|PPC44x_TLB_SW; + /* WIMGE = 0b00100 */ + attrib |= PPC44x_TLB_M; + return attrib; } -- 1.6.0.3