Re: [PATCH] deal with interrupt shadow state for emulated instruction

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Glauber Costa wrote:
we currently unblock shadow interrupt state when we skip an instruction,
but failing to do so when we actually emulate one. This blocks interrupts
in key instruction blocks, in particular sti; hlt; sequences

If the instruction emulated is an sti, we have to block shadow interrupts.
The same goes for mov ss. pop ss also needs it, but we don't currently
emulate it. For sequences of two or more instructions of the same type
among those instructions, only the first one has this effect.

Without this patch, I cannot boot gpxe option roms at vmx machines.
This is described at https://bugzilla.redhat.com/show_bug.cgi?id=494469

--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -513,6 +513,8 @@ struct kvm_x86_ops {
 	void (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
 	int (*handle_exit)(struct kvm_run *run, struct kvm_vcpu *vcpu);
 	void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
+	void (*block_interrupt_shadow)(struct kvm_vcpu *vcpu);
+	void (*unblock_interrupt_shadow)(struct kvm_vcpu *vcpu);

set_interrupt_shadow(), clear_interrupt_shadow(). The interrupt shadow blocks interrupts, but what happens when you block the interrupt shadow?

Maybe better to fold into one callback with a parameter.

 int x86_emulate_insn(struct x86_emulate_ctxt *ctxt,
-		     struct x86_emulate_ops *ops);
+		     struct x86_emulate_ops *ops, int *interruptibility);

Add to x86_emulate_ctxt, there's already some

+static void svm_block_interrupt_shadow(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	svm->vmcb->control.int_state |= SVM_INTERRUPT_SHADOW_MASK;
+	vcpu->arch.interrupt_window_open = 0;
+}
+
+static void svm_unblock_interrupt_shadow(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	svm->vmcb->control.int_state &= ~SVM_INTERRUPT_SHADOW_MASK;
+	vcpu->arch.interrupt_window_open = (svm->vcpu.arch.hflags & HF_GIF_MASK);

If eflags.if = 0, the interrupt window is closed.

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index c6997c0..5158c2b 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -736,26 +736,45 @@ static void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
 	vmcs_writel(GUEST_RFLAGS, rflags);
 }
-static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
+static void vmx_block_interrupt_shadow(struct kvm_vcpu *vcpu)
 {
-	unsigned long rip;
-	u32 interruptibility;
+	u32 interruptibility = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
+	u32 interruptibility_mask = ((GUEST_INTR_STATE_STI | GUEST_INTR_STATE_MOV_SS));
- rip = kvm_rip_read(vcpu);
-	rip += vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
-	kvm_rip_write(vcpu, rip);
+	if (!(interruptibility & interruptibility_mask))
+		vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
+			     interruptibility | interruptibility_mask);
+	vcpu->arch.interrupt_window_open = 0;

Setting both _MOV_SS and _STI is wierd; can't happen on real hardware.

 {
 	unsigned long memop = 0;
 	u64 msr_data;
@@ -1359,6 +1360,10 @@ x86_emulate_insn(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 	unsigned int port;
 	int io_dir_in;
 	int rc = 0;
+	static int movss_int_flag, movss_int_flag_old;

static, for shame.

--
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[Index of Archives]     [KVM ARM]     [KVM ia64]     [KVM ppc]     [Virtualization Tools]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite Questions]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux