Skip to contentSkip to conversation

KVM: Move KVM context switch into own function

PATCH Applied Andi Kleen 3 participants 4 messages Oct 20, 2010 – Oct 21, 2010

Subsystems (3)
lore
Share & tools
lore

Andi Kleen

author
#0 lore raw

PATCH 1/1 KVM: Move KVM context switch into own function

From: Andi Kleen <[email protected]>

gcc 4.5 with some special options is able to duplicate the VMX context switch asm in vmx_vcpu_run(). This results in a compile error because the inline asm sequence uses an on local label. The non local label is needed because other code wants to set up the return address.

This patch moves the asm code into an own function and marks that explicitely noinline to avoid this problem.

Better would be probably to just move it into an .S file.

The diff looks worse than the change really is, it's all just code movement and no logic change.

---
 arch/x86/kvm/vmx.c |   63 +++++++++++++++++++++++++++++++--------------------
 1 files changed, 38 insertions(+), 25 deletions(-)
arch/x86/kvm/vmx.c +38 -25
index 49b25ee..935e33f 100644
@@ -3922,33 +3922,17 @@ static void fixup_rmode_irq(struct vcpu_vmx *vmx)
 #define Q "l"
 #endif
 
-static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
+/* 
+ * We put this into a separate noinline function to prevent the compiler
+ * from duplicating the code. This is needed because this code
+ * uses non local labels that cannot be duplicated.
+ * Do not put any flow control into this function.
+ * Better would be to put this whole monstrosity into a .S file.
+ */
+static void noinline do_vmx_vcpu_run(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
-
-	/* Record the guest's net vcpu time for enforced NMI injections. */
-	if (unlikely(!cpu_has_virtual_nmis() && vmx->soft_vnmi_blocked))
-		vmx->entry_time = ktime_get();
-
-	/* Don't enter VMX if guest state is invalid, let the exit handler
-	   start emulation until we arrive back to a valid state */
-	if (vmx->emulation_required && emulate_invalid_guest_state)
-		return;
-
-	if (test_bit(VCPU_REGS_RSP, (unsigned long *)&vcpu->arch.regs_dirty))
-		vmcs_writel(GUEST_RSP, vcpu->arch.regs[VCPU_REGS_RSP]);
-	if (test_bit(VCPU_REGS_RIP, (unsigned long *)&vcpu->arch.regs_dirty))
-		vmcs_writel(GUEST_RIP, vcpu->arch.regs[VCPU_REGS_RIP]);
-
-	/* When single-stepping over STI and MOV SS, we must clear the
-	 * corresponding interruptibility bits in the guest state. Otherwise
-	 * vmentry fails as it then expects bit 14 (BS) in pending debug
-	 * exceptions being set, but that's not correct for the guest debugging
-	 * case. */
-	if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP)
-		vmx_set_interrupt_shadow(vcpu, 0);
-
-	asm(
+	asm volatile(
 		/* Store host registers */
 		"push %%"R"dx; push %%"R"bp;"
 		"push %%"R"cx \n\t"
@@ -4043,6 +4027,35 @@ static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
 		, "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15"
 #endif
 	      );
+}
+
+static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_vmx *vmx = to_vmx(vcpu);
+
+	/* Record the guest's net vcpu time for enforced NMI injections. */
+	if (unlikely(!cpu_has_virtual_nmis() && vmx->soft_vnmi_blocked))
+		vmx->entry_time = ktime_get();
+
+	/* Don't enter VMX if guest state is invalid, let the exit handler
+	   start emulation until we arrive back to a valid state */
+	if (vmx->emulation_required && emulate_invalid_guest_state)
+		return;
+
+	if (test_bit(VCPU_REGS_RSP, (unsigned long *)&vcpu->arch.regs_dirty))
+		vmcs_writel(GUEST_RSP, vcpu->arch.regs[VCPU_REGS_RSP]);
+	if (test_bit(VCPU_REGS_RIP, (unsigned long *)&vcpu->arch.regs_dirty))
+		vmcs_writel(GUEST_RIP, vcpu->arch.regs[VCPU_REGS_RIP]);
+
+	/* When single-stepping over STI and MOV SS, we must clear the
+	 * corresponding interruptibility bits in the guest state. Otherwise
+	 * vmentry fails as it then expects bit 14 (BS) in pending debug
+	 * exceptions being set, but that's not correct for the guest debugging
+	 * case. */
+	if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP)
+		vmx_set_interrupt_shadow(vcpu, 0);
+
+	do_vmx_vcpu_run(vcpu);
 
 	vcpu->arch.regs_avail = ~((1 << VCPU_REGS_RIP) | (1 << VCPU_REGS_RSP)
 				  | (1 << VCPU_EXREG_PDPTR));
Signature · 1 line
1.7.1
#1 lore raw
Quoted 7 lines · Andi Kleen

From: Andi Kleen<[email protected]>

gcc 4.5 with some special options is able to duplicate the VMX context switch asm in vmx_vcpu_run(). This results in a compile error because the inline asm sequence uses an on local label. The non local label is needed because other code wants to set up the return address.

Curious, any idea why gcc does that? I guess it's the unlikely() near the top.

Anyway, patch looks fine.

Signature · 2 lines
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.
author
#2 lore raw
Quoted 11 lines · Avi Kivity
Quoted 7 lines · Andi Kleen

From: Andi Kleen<[email protected]>

gcc 4.5 with some special options is able to duplicate the VMX context switch asm in vmx_vcpu_run(). This results in a compile error because the inline asm sequence uses an on local label. The non local label is needed because other code wants to set up the return address.

Curious, any idea why gcc does that? I guess it's the unlikely() near the top.

Don't know why, i wasn't motivated to go through its debugging dumps for this. But strictly the problem could always happen, not even a "volatile" stops that.

-Andi

#3 lore raw
In reply to Andi Kleen
Quoted 10 lines · Andi Kleen

From: Andi Kleen <[email protected]>

gcc 4.5 with some special options is able to duplicate the VMX context switch asm in vmx_vcpu_run(). This results in a compile error because the inline asm sequence uses an on local label. The non local label is needed because other code wants to set up the return address.

This patch moves the asm code into an own function and marks that explicitely noinline to avoid this problem.

Quoted 6 lines

Better would be probably to just move it into an .S file.

The diff looks worse than the change really is, it's all just code movement and no logic change.

Applied, thanks.