From mboxrd@z Thu Jan 1 00:00:00 1970 From: Wanpeng Li Subject: Re: [PATCH] KVM: nVMX: Fix vmptrld fail and vmwrite error when L1 goes down w/ enable_shadow_vmcs Date: Tue, 8 Jul 2014 17:27:31 +0800 Message-ID: <20140708092731.GB19860@kernel> References: <1404456758-33691-1-git-send-email-wanpeng.li@linux.intel.com> Reply-To: Wanpeng Li Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Marcelo Tosatti , Bandan Das , kvm@vger.kernel.org, linux-kernel@vger.kernel.org To: Paolo Bonzini , Jan Kiszka Return-path: Content-Disposition: inline In-Reply-To: <1404456758-33691-1-git-send-email-wanpeng.li@linux.intel.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org Ping, On Fri, Jul 04, 2014 at 02:52:38PM +0800, Wanpeng Li wrote: >This bug can be trigger by L1 goes down directly w/ enable_shadow_vmcs. > >[ 6413.158950] kvm: vmptrld (null)/780000000000 failed >[ 6413.158954] vmwrite error: reg 401e value 4 (err 1) >[ 6413.158957] CPU: 0 PID: 4840 Comm: qemu-system-x86 Tainted: G OE 3.16.0kvm+ #2 >[ 6413.158958] Hardware name: Dell Inc. OptiPlex 9020/0DNKMN, BIOS A05 12/05/2013 >[ 6413.158959] 0000000000000003 ffff880210c9fb58 ffffffff81741de9 ffff8800d7433f80 >[ 6413.158960] ffff880210c9fb68 ffffffffa059fa08 ffff880210c9fb78 ffffffffa05938bf >[ 6413.158962] ffff880210c9fba8 ffffffffa059a97f ffff8800d7433f80 0000000000000003 >[ 6413.158963] Call Trace: >[ 6413.158968] [] dump_stack+0x45/0x56 >[ 6413.158972] [] vmwrite_error+0x2c/0x2e [kvm_intel] >[ 6413.158974] [] vmcs_writel+0x1f/0x30 [kvm_intel] >[ 6413.158976] [] free_nested.part.73+0x5f/0x170 [kvm_intel] >[ 6413.158978] [] vmx_free_vcpu+0x33/0x70 [kvm_intel] >[ 6413.158991] [] kvm_arch_vcpu_free+0x44/0x50 [kvm] >[ 6413.158998] [] kvm_arch_destroy_vm+0xf2/0x1f0 [kvm] > >Commit 26a865 (KVM: VMX: fix use after free of vmx->loaded_vmcs) fix the use >after free bug by move free_loaded_vmcs() before free_nested(), however, this >lead to free loaded_vmcs->vmcs premature and vmptrld load a NULL pointer during >sync shadow vmcs to vmcs12. In addition, vmwrite which used to disable shadow >vmcs and reset VMCS_LINK_POINTER failed since there is no valid current-VMCS. >This patch fix it by skipping kfree vmcs02_list item for current-vmcs in >nested_free_all_saved_vmcss() and kfree it after free_loaded_vmcs(). This can >also avoid use after free bug. > >Signed-off-by: Wanpeng Li >--- > arch/x86/kvm/vmx.c | 15 +++++++++++---- > 1 file changed, 11 insertions(+), 4 deletions(-) > >diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c >index 021d84a..6e427ac 100644 >--- a/arch/x86/kvm/vmx.c >+++ b/arch/x86/kvm/vmx.c >@@ -5762,10 +5762,11 @@ static void nested_free_all_saved_vmcss(struct vcpu_vmx *vmx) > { > struct vmcs02_list *item, *n; > list_for_each_entry_safe(item, n, &vmx->nested.vmcs02_pool, list) { >- if (vmx->loaded_vmcs != &item->vmcs02) >+ if (vmx->loaded_vmcs != &item->vmcs02) { > free_loaded_vmcs(&item->vmcs02); >- list_del(&item->list); >- kfree(item); >+ list_del(&item->list); >+ kfree(item); >+ } > } > vmx->nested.vmcs02_num = 0; > >@@ -7526,10 +7527,16 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu) > static void vmx_free_vcpu(struct kvm_vcpu *vcpu) > { > struct vcpu_vmx *vmx = to_vmx(vcpu); >+ struct vmcs02_list *item; > > free_vpid(vmx); >- free_loaded_vmcs(vmx->loaded_vmcs); > free_nested(vmx); >+ free_loaded_vmcs(vmx->loaded_vmcs); >+ if (vmx->nested.vmxon) >+ list_for_each_entry(item, &vmx->nested.vmcs02_pool, list) { >+ list_del(&item->list); >+ kfree(item); >+ } > kfree(vmx->guest_msrs); > kvm_vcpu_uninit(vcpu); > kmem_cache_free(kvm_vcpu_cache, vmx); >-- >1.9.1