KVM: Fix fs/gs reload oops with invalid ldt

kvm reloads the host's fs and gs blindly, however the underlying segment
descriptors may be invalid due to the user modifying the ldt after loading
them.

Fix by using the safe accessors (loadsegment() and load_gs_index()) instead
of home grown unsafe versions.

This is CVE-2010-3698.

KVM-Stable-Tag.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

3 files changed, 19 insertions, 44 deletions

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 502e53f..c52e2eb 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -652,20 +652,6 @@ static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
 	return (struct kvm_mmu_page *)page_private(page);
 }
 
-static inline u16 kvm_read_fs(void)
-{
-	u16 seg;
-	asm("mov %%fs, %0" : "=g"(seg));
-	return seg;
-}
-
-static inline u16 kvm_read_gs(void)
-{
-	u16 seg;
-	asm("mov %%gs, %0" : "=g"(seg));
-	return seg;
-}
-
 static inline u16 kvm_read_ldt(void)
 {
 	u16 ldt;
@@ -673,16 +659,6 @@ static inline u16 kvm_read_ldt(void)
 	return ldt;
 }
 
-static inline void kvm_load_fs(u16 sel)
-{
-	asm("mov %0, %%fs" : : "rm"(sel));
-}
-
-static inline void kvm_load_gs(u16 sel)
-{
-	asm("mov %0, %%gs" : : "rm"(sel));
-}
-
 static inline void kvm_load_ldt(u16 sel)
 {
 	asm("lldt %0" : : "rm"(sel));
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 81ed28c..8a3f9f6 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -3163,8 +3163,8 @@ static void svm_vcpu_run(struct kvm_vcpu *vcpu)
 	sync_lapic_to_cr8(vcpu);
 
 	save_host_msrs(vcpu);
-	fs_selector = kvm_read_fs();
-	gs_selector = kvm_read_gs();
+	savesegment(fs, fs_selector);
+	savesegment(gs, gs_selector);
 	ldt_selector = kvm_read_ldt();
 	svm->vmcb->save.cr2 = vcpu->arch.cr2;
 	/* required for live migration with NPT */
@@ -3251,10 +3251,15 @@ static void svm_vcpu_run(struct kvm_vcpu *vcpu)
 	vcpu->arch.regs[VCPU_REGS_RSP] = svm->vmcb->save.rsp;
 	vcpu->arch.regs[VCPU_REGS_RIP] = svm->vmcb->save.rip;
 
-	kvm_load_fs(fs_selector);
-	kvm_load_gs(gs_selector);
-	kvm_load_ldt(ldt_selector);
 	load_host_msrs(vcpu);
+	loadsegment(fs, fs_selector);
+#ifdef CONFIG_X86_64
+	load_gs_index(gs_selector);
+	wrmsrl(MSR_KERNEL_GS_BASE, current->thread.gs);
+#else
+	loadsegment(gs, gs_selector);
+#endif
+	kvm_load_ldt(ldt_selector);
 
 	reload_tss(vcpu);
 
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 49b25ee..7bddfab 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -803,7 +803,7 @@ static void vmx_save_host_state(struct kvm_vcpu *vcpu)
 	 */
 	vmx->host_state.ldt_sel = kvm_read_ldt();
 	vmx->host_state.gs_ldt_reload_needed = vmx->host_state.ldt_sel;
-	vmx->host_state.fs_sel = kvm_read_fs();
+	savesegment(fs, vmx->host_state.fs_sel);
 	if (!(vmx->host_state.fs_sel & 7)) {
 		vmcs_write16(HOST_FS_SELECTOR, vmx->host_state.fs_sel);
 		vmx->host_state.fs_reload_needed = 0;
@@ -811,7 +811,7 @@ static void vmx_save_host_state(struct kvm_vcpu *vcpu)
 		vmcs_write16(HOST_FS_SELECTOR, 0);
 		vmx->host_state.fs_reload_needed = 1;
 	}
-	vmx->host_state.gs_sel = kvm_read_gs();
+	savesegment(gs, vmx->host_state.gs_sel);
 	if (!(vmx->host_state.gs_sel & 7))
 		vmcs_write16(HOST_GS_SELECTOR, vmx->host_state.gs_sel);
 	else {
@@ -841,27 +841,21 @@ static void vmx_save_host_state(struct kvm_vcpu *vcpu)
 
 static void __vmx_load_host_state(struct vcpu_vmx *vmx)
 {
-	unsigned long flags;
-
 	if (!vmx->host_state.loaded)
 		return;
 
 	++vmx->vcpu.stat.host_state_reload;
 	vmx->host_state.loaded = 0;
 	if (vmx->host_state.fs_reload_needed)
-		kvm_load_fs(vmx->host_state.fs_sel);
+		loadsegment(fs, vmx->host_state.fs_sel);
 	if (vmx->host_state.gs_ldt_reload_needed) {
 		kvm_load_ldt(vmx->host_state.ldt_sel);
-		/*
-		 * If we have to reload gs, we must take care to
-		 * preserve our gs base.
-		 */
-		local_irq_save(flags);
-		kvm_load_gs(vmx->host_state.gs_sel);
 #ifdef CONFIG_X86_64
-		wrmsrl(MSR_GS_BASE, vmcs_readl(HOST_GS_BASE));
+		load_gs_index(vmx->host_state.gs_sel);
+		wrmsrl(MSR_KERNEL_GS_BASE, current->thread.gs);
+#else
+		loadsegment(gs, vmx->host_state.gs_sel);
 #endif
-		local_irq_restore(flags);
 	}
 	reload_tss();
 #ifdef CONFIG_X86_64
@@ -2589,8 +2583,8 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
 	vmcs_write16(HOST_CS_SELECTOR, __KERNEL_CS);  /* 22.2.4 */
 	vmcs_write16(HOST_DS_SELECTOR, __KERNEL_DS);  /* 22.2.4 */
 	vmcs_write16(HOST_ES_SELECTOR, __KERNEL_DS);  /* 22.2.4 */
-	vmcs_write16(HOST_FS_SELECTOR, kvm_read_fs());    /* 22.2.4 */
-	vmcs_write16(HOST_GS_SELECTOR, kvm_read_gs());    /* 22.2.4 */
+	vmcs_write16(HOST_FS_SELECTOR, 0);            /* 22.2.4 */
+	vmcs_write16(HOST_GS_SELECTOR, 0);            /* 22.2.4 */
 	vmcs_write16(HOST_SS_SELECTOR, __KERNEL_DS);  /* 22.2.4 */
 #ifdef CONFIG_X86_64
 	rdmsrl(MSR_FS_BASE, a);