MSI-X MMIO support for KVM

Add a new parameter to IO writing handler, so that we can transfer information from IO handler to caller.

Signed-off-by Sheng Yang <[email protected]>

---
 arch/x86/kvm/i8254.c      |    6 ++++--
 arch/x86/kvm/i8259.c      |    3 ++-
 arch/x86/kvm/lapic.c      |    3 ++-
 arch/x86/kvm/x86.c        |   13 ++++++++-----
 include/linux/kvm_host.h  |   12 ++++++++++--
 virt/kvm/coalesced_mmio.c |    3 ++-
 virt/kvm/eventfd.c        |    2 +-
 virt/kvm/ioapic.c         |    2 +-
 virt/kvm/iodev.h          |    6 ++++--
 virt/kvm/kvm_main.c       |    4 ++--
 10 files changed, 36 insertions(+), 18 deletions(-)

arch/x86/kvm/i8254.c +4 -2

index efad723..bd8f0c5 100644
@@ -439,7 +439,8 @@ static inline int pit_in_range(gpa_t addr)
 }
 
 static int pit_ioport_write(struct kvm_io_device *this,
-			    gpa_t addr, int len, const void *data)
+			    gpa_t addr, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)
 {
 	struct kvm_pit *pit = dev_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

@@ -585,7 +586,8 @@ static int pit_ioport_read(struct kvm_io_device *this,
 }
 
 static int speaker_ioport_write(struct kvm_io_device *this,
-				gpa_t addr, int len, const void *data)
+				gpa_t addr, int len, const void *data,
+				struct kvm_io_ext_data *ext_data)
 {
 	struct kvm_pit *pit = speaker_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

arch/x86/kvm/i8259.c +2 -1

index 3cece05..96b1070 100644
@@ -480,7 +480,8 @@ static inline struct kvm_pic *to_pic(struct kvm_io_device *dev)
 }
 
 static int picdev_write(struct kvm_io_device *this,
-			 gpa_t addr, int len, const void *val)
+			 gpa_t addr, int len, const void *val,
+			 struct kvm_io_ext_data *ext_data)
 {
 	struct kvm_pic *s = to_pic(this);
 	unsigned char data = *(unsigned char *)val;

arch/x86/kvm/lapic.c +2 -1

index 93cf9d0..f413e9c 100644
@@ -836,7 +836,8 @@ static int apic_reg_write(struct kvm_lapic *apic, u32 reg, u32 val)
 }
 
 static int apic_mmio_write(struct kvm_io_device *this,
-			    gpa_t address, int len, const void *data)
+			    gpa_t address, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)
 {
 	struct kvm_lapic *apic = to_lapic(this);
 	unsigned int offset = address - apic->base_address;

arch/x86/kvm/x86.c +8 -5

index fa708c9..21b84e2 100644
@@ -3571,13 +3571,14 @@ static void kvm_init_msr_list(void)
 }
 
 static int vcpu_mmio_write(struct kvm_vcpu *vcpu, gpa_t addr, int len,
-			   const void *v)
+			   const void *v, struct kvm_io_ext_data *ext_data)
 {
 	if (vcpu->arch.apic &&
-	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v))
+	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v, ext_data))
 		return 0;
 
-	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS, addr, len, v);
+	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS,
+				addr, len, v, ext_data);
 }
 
 static int vcpu_mmio_read(struct kvm_vcpu *vcpu, gpa_t addr, int len, void *v)

@@ -3807,6 +3808,7 @@ static int emulator_write_emulated_onepage(unsigned long addr,
 					   struct kvm_vcpu *vcpu)
 {
 	gpa_t                 gpa;
+	struct kvm_io_ext_data ext_data;
 
 	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, exception);
 

@@ -3825,7 +3827,7 @@ mmio:
 	/*
 	 * Is this MMIO handled locally?
 	 */
-	if (!vcpu_mmio_write(vcpu, gpa, bytes, val))
+	if (!vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data))
 		return X86EMUL_CONTINUE;
 
 	vcpu->mmio_needed = 1;

@@ -3940,6 +3942,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu, void *pd)
 {
 	/* TODO: String I/O for in kernel device */
 	int r;
+	struct kvm_io_ext_data ext_data;
 
 	if (vcpu->arch.pio.in)
 		r = kvm_io_bus_read(vcpu->kvm, KVM_PIO_BUS, vcpu->arch.pio.port,

@@ -3947,7 +3950,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu, void *pd)
 	else
 		r = kvm_io_bus_write(vcpu->kvm, KVM_PIO_BUS,
 				     vcpu->arch.pio.port, vcpu->arch.pio.size,
-				     pd);
+				     pd, &ext_data);
 	return r;
 }
 

include/linux/kvm_host.h +0 -0

index 7d313e0..6bb211d 100644

include/linux/kvm_host.h +5 -2

+};
+
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
-		     int len, const void *val);
+		     int len, const void *val, struct kvm_io_ext_data *data);
 int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr, int len,
 		    void *val);
 int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx,

@@ -113,7 +120,8 @@ struct kvm_io_device_ops {
 	int (*write)(struct kvm_io_device *this,
 		     gpa_t addr,
 		     int len,
-		     const void *val);
+		     const void *val,
+		     struct kvm_io_ext_data *data);
 	void (*destructor)(struct kvm_io_device *this);
 };
 

virt/kvm/coalesced_mmio.c +2 -1

index fc84875..37b254c 100644
@@ -59,7 +59,8 @@ static int coalesced_mmio_in_range(struct kvm_coalesced_mmio_dev *dev,
 }
 
 static int coalesced_mmio_write(struct kvm_io_device *this,
-				gpa_t addr, int len, const void *val)
+				gpa_t addr, int len, const void *val,
+				struct kvm_io_ext_data *ext_data)
 {
 	struct kvm_coalesced_mmio_dev *dev = to_mmio(this);
 	struct kvm_coalesced_mmio_ring *ring = dev->kvm->coalesced_mmio_ring;

virt/kvm/eventfd.c +1 -1

index 2ca4535..8edd757 100644
@@ -483,7 +483,7 @@ ioeventfd_in_range(struct _ioeventfd *p, gpa_t addr, int len, const void *val)
 /* MMIO/PIO writes trigger an event if the addr/val match */
 static int
 ioeventfd_write(struct kvm_io_device *this, gpa_t addr, int len,
-		const void *val)
+		const void *val, struct kvm_io_ext_data *ext_data)
 {
 	struct _ioeventfd *p = to_ioeventfd(this);
 

virt/kvm/ioapic.c +1 -1

index 0b9df83..6a027ef 100644
@@ -321,7 +321,7 @@ static int ioapic_mmio_read(struct kvm_io_device *this, gpa_t addr, int len,
 }
 
 static int ioapic_mmio_write(struct kvm_io_device *this, gpa_t addr, int len,
-			     const void *val)
+			     const void *val, struct kvm_io_ext_data *ext_data)
 {
 	struct kvm_ioapic *ioapic = to_ioapic(this);
 	u32 data;

virt/kvm/iodev.h +4 -2

index d1f5651..340ab79 100644
@@ -33,9 +33,11 @@ static inline int kvm_iodevice_read(struct kvm_io_device *dev,
 }
 
 static inline int kvm_iodevice_write(struct kvm_io_device *dev,
-				     gpa_t addr, int l, const void *v)
+				     gpa_t addr, int l, const void *v,
+				     struct kvm_io_ext_data *data)
 {
-	return dev->ops->write ? dev->ops->write(dev, addr, l, v) : -EOPNOTSUPP;
+	return dev->ops->write ?
+		dev->ops->write(dev, addr, l, v, data) : -EOPNOTSUPP;
 }
 
 static inline void kvm_iodevice_destructor(struct kvm_io_device *dev)

virt/kvm/kvm_main.c +2 -2

index b1b6cbb..a61f90e 100644
@@ -2221,14 +2221,14 @@ static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
 
 /* kvm_io_bus_write - called under kvm->slots_lock */
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
-		     int len, const void *val)
+		     int len, const void *val, struct kvm_io_ext_data *ext_data)
 {
 	int i;
 	struct kvm_io_bus *bus;
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
 	for (i = 0; i < bus->dev_count; i++)
-		if (!kvm_iodevice_write(bus->devs[i], addr, len, val))
+		if (!kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data))
 			return 0;
 	return -EOPNOTSUPP;
 }

Signature · 1 line

1.7.0.1

Quoted 186 lines · Sheng Yang

Add a new parameter to IO writing handler, so that we can transfer information from IO handler to caller.

Signed-off-by Sheng Yang <[email protected]>

---

 arch/x86/kvm/i8254.c      |    6 ++++--
 arch/x86/kvm/i8259.c      |    3 ++-
 arch/x86/kvm/lapic.c      |    3 ++-
 arch/x86/kvm/x86.c        |   13 ++++++++-----
 include/linux/kvm_host.h  |   12 ++++++++++--
 virt/kvm/coalesced_mmio.c |    3 ++-
 virt/kvm/eventfd.c        |    2 +-
 virt/kvm/ioapic.c         |    2 +-
 virt/kvm/iodev.h          |    6 ++++--
 virt/kvm/kvm_main.c       |    4 ++--
 10 files changed, 36 insertions(+), 18 deletions(-)

arch/x86/kvm/i8254.c +2 -1

index efad723..bd8f0c5 100644
@@ -439,7 +439,8 @@ static inline int pit_in_range(gpa_t addr)
 
 }
 
 static int pit_ioport_write(struct kvm_io_device *this,
 
-			    gpa_t addr, int len, const void *data)
+			    gpa_t addr, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)
 

{

arch/x86/kvm/i8254.c +2 -1

 	struct kvm_pit *pit = dev_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

@@ -585,7 +586,8 @@ static int pit_ioport_read(struct kvm_io_device
 *this,
 
 }
 
 static int speaker_ioport_write(struct kvm_io_device *this,
 
-				gpa_t addr, int len, const void *data)
+				gpa_t addr, int len, const void *data,
+				struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/i8254.c +0 -0

 	struct kvm_pit *pit = speaker_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

arch/x86/kvm/i8259.c +2 -1

index 3cece05..96b1070 100644
@@ -480,7 +480,8 @@ static inline struct kvm_pic *to_pic(struct
 kvm_io_device *dev)
 
 }
 
 static int picdev_write(struct kvm_io_device *this,
 
-			 gpa_t addr, int len, const void *val)
+			 gpa_t addr, int len, const void *val,
+			 struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/i8259.c +0 -0

 	struct kvm_pic *s = to_pic(this);
 	unsigned char data = *(unsigned char *)val;

arch/x86/kvm/lapic.c +2 -1

index 93cf9d0..f413e9c 100644
@@ -836,7 +836,8 @@ static int apic_reg_write(struct kvm_lapic *apic, u32
 reg, u32 val)
 
 }
 
 static int apic_mmio_write(struct kvm_io_device *this,
 
-			    gpa_t address, int len, const void *data)
+			    gpa_t address, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/lapic.c +0 -0

 	struct kvm_lapic *apic = to_lapic(this);
 	unsigned int offset = address - apic->base_address;

arch/x86/kvm/x86.c +6 -4

index fa708c9..21b84e2 100644
@@ -3571,13 +3571,14 @@ static void kvm_init_msr_list(void)
 
 }
 
 static int vcpu_mmio_write(struct kvm_vcpu *vcpu, gpa_t addr, int len,
 
-			   const void *v)
+			   const void *v, struct kvm_io_ext_data *ext_data)
 
 {
 
 	if (vcpu->arch.apic &&
 
-	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v))
+	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v, ext_data))
 
 		return 0;
 
-	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS, addr, len, v);
+	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS,
+				addr, len, v, ext_data);
 
 }
 
 static int vcpu_mmio_read(struct kvm_vcpu *vcpu, gpa_t addr, int len,
 void *v)

@@ -3807,6 +3808,7 @@ static int emulator_write_emulated_onepage(unsigned
 long addr,
 
 					   struct kvm_vcpu *vcpu)
 
 {
 
 	gpa_t                 gpa;
 
+	struct kvm_io_ext_data ext_data;
 
 	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, exception);

@@ -3825,7 +3827,7 @@ mmio:
 	/*
 
 	 * Is this MMIO handled locally?
 	 */
 
-	if (!vcpu_mmio_write(vcpu, gpa, bytes, val))
+	if (!vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data))
 

arch/x86/kvm/x86.c +2 -1

 		return X86EMUL_CONTINUE;
 
 	vcpu->mmio_needed = 1;

@@ -3940,6 +3942,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu, void
 *pd)
 
 {
 
 	/* TODO: String I/O for in kernel device */
 	int r;
 
+	struct kvm_io_ext_data ext_data;
 
 	if (vcpu->arch.pio.in)
 
 		r = kvm_io_bus_read(vcpu->kvm, KVM_PIO_BUS, vcpu->arch.pio.port,

@@ -3947,7 +3950,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu, void
 *pd)
 
 	else
 
 		r = kvm_io_bus_write(vcpu->kvm, KVM_PIO_BUS,
 
 				     vcpu->arch.pio.port, vcpu->arch.pio.size,
 
-				     pd);
+				     pd, &ext_data);
 
 	return r;
 
 }

include/linux/kvm_host.h +0 -0

index 7d313e0..6bb211d 100644
@@ -68,8 +68,15 @@ enum kvm_bus {
 

include/linux/kvm_host.h enum kvm_bus { · line 72

Quoted 6 lines

include/linux/kvm_host.h +2 -0

 	KVM_NR_BUSES
 
 };
 
+struct kvm_io_ext_data {
+	int type;

What values does this get? Please add documentation in comments.

include/linux/kvm_host.h enum kvm_bus { · line 75

Quoted 4 lines

include/linux/kvm_host.h +3 -0

+	union {
+		char padding[256];
+	};

So the structure size is 260 bytes?
What's the point of the padding?

Quoted 143 lines

include/linux/kvm_host.h +5 -2

+};
+
 
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 
-		     int len, const void *val);
+		     int len, const void *val, struct kvm_io_ext_data *data);
 
 int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 int len,
 
 		    void *val);
 
 int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx,

@@ -113,7 +120,8 @@ struct kvm_io_device_ops {
 
 	int (*write)(struct kvm_io_device *this,
 
 		     gpa_t addr,
 		     int len,
 
-		     const void *val);
+		     const void *val,
+		     struct kvm_io_ext_data *data);

include/linux/kvm_host.h +0 -0

 	void (*destructor)(struct kvm_io_device *this);
 
 };

virt/kvm/coalesced_mmio.c +2 -1

index fc84875..37b254c 100644
@@ -59,7 +59,8 @@ static int coalesced_mmio_in_range(struct
 kvm_coalesced_mmio_dev *dev,
 
 }
 
 static int coalesced_mmio_write(struct kvm_io_device *this,
 
-				gpa_t addr, int len, const void *val)
+				gpa_t addr, int len, const void *val,
+				struct kvm_io_ext_data *ext_data)

{

virt/kvm/coalesced_mmio.c +0 -0

 	struct kvm_coalesced_mmio_dev *dev = to_mmio(this);
 	struct kvm_coalesced_mmio_ring *ring = dev->kvm->coalesced_mmio_ring;

virt/kvm/eventfd.c +1 -1

index 2ca4535..8edd757 100644
@@ -483,7 +483,7 @@ ioeventfd_in_range(struct _ioeventfd *p, gpa_t addr,
 int len, const void *val)
 
 /* MMIO/PIO writes trigger an event if the addr/val match */
 static int
 ioeventfd_write(struct kvm_io_device *this, gpa_t addr, int len,
 
-		const void *val)
+		const void *val, struct kvm_io_ext_data *ext_data)

{

virt/kvm/eventfd.c +0 -0

 	struct _ioeventfd *p = to_ioeventfd(this);

virt/kvm/ioapic.c +1 -1

index 0b9df83..6a027ef 100644
@@ -321,7 +321,7 @@ static int ioapic_mmio_read(struct kvm_io_device
 *this, gpa_t addr, int len,
 
 }
 
 static int ioapic_mmio_write(struct kvm_io_device *this, gpa_t addr, int
 len,
 
-			     const void *val)
+			     const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
 	struct kvm_ioapic *ioapic = to_ioapic(this);
 	u32 data;

virt/kvm/iodev.h +3 -3

index d1f5651..340ab79 100644
@@ -33,9 +33,11 @@ static inline int kvm_iodevice_read(struct
 kvm_io_device *dev,
 
 }
 
 static inline int kvm_iodevice_write(struct kvm_io_device *dev,
 
-				     gpa_t addr, int l, const void *v)
+				     gpa_t addr, int l, const void *v,
+				     struct kvm_io_ext_data *data)
 
 {
 
-	return dev->ops->write ? dev->ops->write(dev, addr, l, v) :
-EOPNOTSUPP; +	return dev->ops->write ?
+		dev->ops->write(dev, addr, l, v, data) : -EOPNOTSUPP;
 
 }
 
 static inline void kvm_iodevice_destructor(struct kvm_io_device *dev)

virt/kvm/kvm_main.c +2 -2

index b1b6cbb..a61f90e 100644
@@ -2221,14 +2221,14 @@ static void kvm_io_bus_destroy(struct kvm_io_bus
 *bus)
 
 /* kvm_io_bus_write - called under kvm->slots_lock */
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 
-		     int len, const void *val)
+		     int len, const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
 	int i;
 	struct kvm_io_bus *bus;
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
 	for (i = 0; i < bus->dev_count; i++)
 
-		if (!kvm_iodevice_write(bus->devs[i], addr, len, val))
+		if (!kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data))
 
 			return 0;
 
 	return -EOPNOTSUPP;
 
 }

Quoted 187 lines · On Thursday 24 February 2011 18:22:19 Michael S. Tsirkin wrote:

Quoted 186 lines · Sheng Yang

Add a new parameter to IO writing handler, so that we can transfer information from IO handler to caller.

Signed-off-by Sheng Yang <[email protected]>

---

 arch/x86/kvm/i8254.c      |    6 ++++--
 arch/x86/kvm/i8259.c      |    3 ++-
 arch/x86/kvm/lapic.c      |    3 ++-
 arch/x86/kvm/x86.c        |   13 ++++++++-----
 include/linux/kvm_host.h  |   12 ++++++++++--
 virt/kvm/coalesced_mmio.c |    3 ++-
 virt/kvm/eventfd.c        |    2 +-
 virt/kvm/ioapic.c         |    2 +-
 virt/kvm/iodev.h          |    6 ++++--
 virt/kvm/kvm_main.c       |    4 ++--
 10 files changed, 36 insertions(+), 18 deletions(-)

arch/x86/kvm/i8254.c +2 -1

index efad723..bd8f0c5 100644
@@ -439,7 +439,8 @@ static inline int pit_in_range(gpa_t addr)
 
 }
 
 static int pit_ioport_write(struct kvm_io_device *this,
 
-			    gpa_t addr, int len, const void *data)
+			    gpa_t addr, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)
 

{

arch/x86/kvm/i8254.c +2 -1

 	struct kvm_pit *pit = dev_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

@@ -585,7 +586,8 @@ static int pit_ioport_read(struct kvm_io_device
 *this,
 
 }
 
 static int speaker_ioport_write(struct kvm_io_device *this,
 
-				gpa_t addr, int len, const void *data)
+				gpa_t addr, int len, const void *data,
+				struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/i8254.c +0 -0

 	struct kvm_pit *pit = speaker_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

arch/x86/kvm/i8259.c +2 -1

index 3cece05..96b1070 100644
@@ -480,7 +480,8 @@ static inline struct kvm_pic *to_pic(struct
 kvm_io_device *dev)
 
 }
 
 static int picdev_write(struct kvm_io_device *this,
 
-			 gpa_t addr, int len, const void *val)
+			 gpa_t addr, int len, const void *val,
+			 struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/i8259.c +0 -0

 	struct kvm_pic *s = to_pic(this);
 	unsigned char data = *(unsigned char *)val;

arch/x86/kvm/lapic.c +2 -1

index 93cf9d0..f413e9c 100644
@@ -836,7 +836,8 @@ static int apic_reg_write(struct kvm_lapic *apic, u32
 reg, u32 val)
 
 }
 
 static int apic_mmio_write(struct kvm_io_device *this,
 
-			    gpa_t address, int len, const void *data)
+			    gpa_t address, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/lapic.c +0 -0

 	struct kvm_lapic *apic = to_lapic(this);
 	unsigned int offset = address - apic->base_address;

arch/x86/kvm/x86.c +6 -4

index fa708c9..21b84e2 100644
@@ -3571,13 +3571,14 @@ static void kvm_init_msr_list(void)
 
 }
 
 static int vcpu_mmio_write(struct kvm_vcpu *vcpu, gpa_t addr, int len,
 
-			   const void *v)
+			   const void *v, struct kvm_io_ext_data *ext_data)
 
 {
 
 	if (vcpu->arch.apic &&
 
-	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v))
+	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v, ext_data))
 
 		return 0;
 
-	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS, addr, len, v);
+	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS,
+				addr, len, v, ext_data);
 
 }
 
 static int vcpu_mmio_read(struct kvm_vcpu *vcpu, gpa_t addr, int len,
 void *v)

@@ -3807,6 +3808,7 @@ static int emulator_write_emulated_onepage(unsigned
 long addr,
 
 					   struct kvm_vcpu *vcpu)
 
 {
 
 	gpa_t                 gpa;
 
+	struct kvm_io_ext_data ext_data;
 
 	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, exception);

@@ -3825,7 +3827,7 @@ mmio:
 	/*
 
 	 * Is this MMIO handled locally?
 	 */
 
-	if (!vcpu_mmio_write(vcpu, gpa, bytes, val))
+	if (!vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data))
 

arch/x86/kvm/x86.c +2 -1

 		return X86EMUL_CONTINUE;
 
 	vcpu->mmio_needed = 1;

@@ -3940,6 +3942,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu, void
 *pd)
 
 {
 
 	/* TODO: String I/O for in kernel device */
 	int r;
 
+	struct kvm_io_ext_data ext_data;
 
 	if (vcpu->arch.pio.in)
 
 		r = kvm_io_bus_read(vcpu->kvm, KVM_PIO_BUS, vcpu->arch.pio.port,

@@ -3947,7 +3950,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu, void
 *pd)
 
 	else
 
 		r = kvm_io_bus_write(vcpu->kvm, KVM_PIO_BUS,
 
 				     vcpu->arch.pio.port, vcpu->arch.pio.size,
 
-				     pd);
+				     pd, &ext_data);
 
 	return r;
 
 }

include/linux/kvm_host.h +0 -0

index 7d313e0..6bb211d 100644
@@ -68,8 +68,15 @@ enum kvm_bus {
 

Quoted 8 lines

include/linux/kvm_host.h enum kvm_bus { · line 72

Quoted 6 lines

include/linux/kvm_host.h +2 -0

 	KVM_NR_BUSES
 
 };
 
+struct kvm_io_ext_data {
+	int type;

What values does this get? Please add documentation in comments.

See the next patch.

Quoted 1 line

Quoted 6 lines

include/linux/kvm_host.h enum kvm_bus { · line 75

Quoted 3 lines

include/linux/kvm_host.h +3 -0

+	union {
+		char padding[256];
+	};

So the structure size is 260 bytes?
What's the point of the padding?

Reserved spaces. Also used in the next patch.

--
regards
Yang, Sheng

Quoted 143 lines

Quoted 143 lines

include/linux/kvm_host.h +5 -2

+};
+
 
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 
-		     int len, const void *val);
+		     int len, const void *val, struct kvm_io_ext_data *data);
 
 int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 int len,
 
 		    void *val);
 
 int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx,

@@ -113,7 +120,8 @@ struct kvm_io_device_ops {
 
 	int (*write)(struct kvm_io_device *this,
 
 		     gpa_t addr,
 		     int len,
 
-		     const void *val);
+		     const void *val,
+		     struct kvm_io_ext_data *data);

include/linux/kvm_host.h +0 -0

 	void (*destructor)(struct kvm_io_device *this);
 
 };

virt/kvm/coalesced_mmio.c +2 -1

index fc84875..37b254c 100644
@@ -59,7 +59,8 @@ static int coalesced_mmio_in_range(struct
 kvm_coalesced_mmio_dev *dev,
 
 }
 
 static int coalesced_mmio_write(struct kvm_io_device *this,
 
-				gpa_t addr, int len, const void *val)
+				gpa_t addr, int len, const void *val,
+				struct kvm_io_ext_data *ext_data)

{

virt/kvm/coalesced_mmio.c +0 -0

 	struct kvm_coalesced_mmio_dev *dev = to_mmio(this);
 	struct kvm_coalesced_mmio_ring *ring = dev->kvm->coalesced_mmio_ring;

virt/kvm/eventfd.c +1 -1

index 2ca4535..8edd757 100644
@@ -483,7 +483,7 @@ ioeventfd_in_range(struct _ioeventfd *p, gpa_t addr,
 int len, const void *val)
 
 /* MMIO/PIO writes trigger an event if the addr/val match */
 static int
 ioeventfd_write(struct kvm_io_device *this, gpa_t addr, int len,
 
-		const void *val)
+		const void *val, struct kvm_io_ext_data *ext_data)

{

virt/kvm/eventfd.c +0 -0

 	struct _ioeventfd *p = to_ioeventfd(this);

virt/kvm/ioapic.c +1 -1

index 0b9df83..6a027ef 100644
@@ -321,7 +321,7 @@ static int ioapic_mmio_read(struct kvm_io_device
 *this, gpa_t addr, int len,
 
 }
 
 static int ioapic_mmio_write(struct kvm_io_device *this, gpa_t addr, int
 len,
 
-			     const void *val)
+			     const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
 	struct kvm_ioapic *ioapic = to_ioapic(this);
 	u32 data;

virt/kvm/iodev.h +3 -3

index d1f5651..340ab79 100644
@@ -33,9 +33,11 @@ static inline int kvm_iodevice_read(struct
 kvm_io_device *dev,
 
 }
 
 static inline int kvm_iodevice_write(struct kvm_io_device *dev,
 
-				     gpa_t addr, int l, const void *v)
+				     gpa_t addr, int l, const void *v,
+				     struct kvm_io_ext_data *data)
 
 {
 
-	return dev->ops->write ? dev->ops->write(dev, addr, l, v) :
-EOPNOTSUPP; +	return dev->ops->write ?
+		dev->ops->write(dev, addr, l, v, data) : -EOPNOTSUPP;
 
 }
 
 static inline void kvm_iodevice_destructor(struct kvm_io_device *dev)

virt/kvm/kvm_main.c +2 -2

index b1b6cbb..a61f90e 100644
@@ -2221,14 +2221,14 @@ static void kvm_io_bus_destroy(struct kvm_io_bus
 *bus)
 
 /* kvm_io_bus_write - called under kvm->slots_lock */
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 
-		     int len, const void *val)
+		     int len, const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
 	int i;
 	struct kvm_io_bus *bus;
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
 	for (i = 0; i < bus->dev_count; i++)
 
-		if (!kvm_iodevice_write(bus->devs[i], addr, len, val))
+		if (!kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data))
 
 			return 0;
 
 	return -EOPNOTSUPP;
 
 }

Quoted 159 lines · Sheng Yang

Quoted 158 lines · On Thursday 24 February 2011 18:22:19 Michael S. Tsirkin wrote:

Quoted 157 lines · Sheng Yang

Add a new parameter to IO writing handler, so that we can transfer information from IO handler to caller.

Signed-off-by Sheng Yang <[email protected]>

---

 arch/x86/kvm/i8254.c      |    6 ++++--
 arch/x86/kvm/i8259.c      |    3 ++-
 arch/x86/kvm/lapic.c      |    3 ++-
 arch/x86/kvm/x86.c        |   13 ++++++++-----
 include/linux/kvm_host.h  |   12 ++++++++++--
 virt/kvm/coalesced_mmio.c |    3 ++-
 virt/kvm/eventfd.c        |    2 +-
 virt/kvm/ioapic.c         |    2 +-
 virt/kvm/iodev.h          |    6 ++++--
 virt/kvm/kvm_main.c       |    4 ++--
 10 files changed, 36 insertions(+), 18 deletions(-)

arch/x86/kvm/i8254.c +2 -1

index efad723..bd8f0c5 100644
@@ -439,7 +439,8 @@ static inline int pit_in_range(gpa_t addr)
 
 }
 
 static int pit_ioport_write(struct kvm_io_device *this,
 
-			    gpa_t addr, int len, const void *data)
+			    gpa_t addr, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)
 

{

arch/x86/kvm/i8254.c +2 -1

 	struct kvm_pit *pit = dev_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

@@ -585,7 +586,8 @@ static int pit_ioport_read(struct kvm_io_device
 *this,
 
 }
 
 static int speaker_ioport_write(struct kvm_io_device *this,
 
-				gpa_t addr, int len, const void *data)
+				gpa_t addr, int len, const void *data,
+				struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/i8254.c +0 -0

 	struct kvm_pit *pit = speaker_to_pit(this);
 	struct kvm_kpit_state *pit_state = &pit->pit_state;

arch/x86/kvm/i8259.c +2 -1

index 3cece05..96b1070 100644
@@ -480,7 +480,8 @@ static inline struct kvm_pic *to_pic(struct
 kvm_io_device *dev)
 
 }
 
 static int picdev_write(struct kvm_io_device *this,
 
-			 gpa_t addr, int len, const void *val)
+			 gpa_t addr, int len, const void *val,
+			 struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/i8259.c +0 -0

 	struct kvm_pic *s = to_pic(this);
 	unsigned char data = *(unsigned char *)val;

arch/x86/kvm/lapic.c +2 -1

index 93cf9d0..f413e9c 100644
@@ -836,7 +836,8 @@ static int apic_reg_write(struct kvm_lapic *apic,
 u32 reg, u32 val)
 
 }
 
 static int apic_mmio_write(struct kvm_io_device *this,
 
-			    gpa_t address, int len, const void *data)
+			    gpa_t address, int len, const void *data,
+			    struct kvm_io_ext_data *ext_data)

{

arch/x86/kvm/lapic.c +0 -0

 	struct kvm_lapic *apic = to_lapic(this);
 	unsigned int offset = address - apic->base_address;

arch/x86/kvm/x86.c +2 -2

index fa708c9..21b84e2 100644
@@ -3571,13 +3571,14 @@ static void kvm_init_msr_list(void)
 
 }
 
 static int vcpu_mmio_write(struct kvm_vcpu *vcpu, gpa_t addr, int
 len,
 
-			   const void *v)
+			   const void *v, struct kvm_io_ext_data *ext_data)
 
 {
 
 	if (vcpu->arch.apic &&
 
-	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v))
+	    !kvm_iodevice_write(&vcpu->arch.apic->dev, addr, len, v,

ext_data))

arch/x86/kvm/x86.c +4 -2

 		return 0;
 
-	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS, addr, len, v);
+	return kvm_io_bus_write(vcpu->kvm, KVM_MMIO_BUS,
+				addr, len, v, ext_data);
 
 }
 
 static int vcpu_mmio_read(struct kvm_vcpu *vcpu, gpa_t addr, int
 len, void *v)

@@ -3807,6 +3808,7 @@ static int
 emulator_write_emulated_onepage(unsigned long addr,
 
 					   struct kvm_vcpu *vcpu)
 
 {
 
 	gpa_t                 gpa;
 
+	struct kvm_io_ext_data ext_data;
 
 	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, exception);

@@ -3825,7 +3827,7 @@ mmio:
 	/*
 
 	 * Is this MMIO handled locally?
 	 */
 
-	if (!vcpu_mmio_write(vcpu, gpa, bytes, val))
+	if (!vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data))
 

arch/x86/kvm/x86.c +0 -0

 		return X86EMUL_CONTINUE;
 
 	vcpu->mmio_needed = 1;

@@ -3940,6 +3942,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu,
 void *pd)
 
 {
 

Quoted 8 lines

Quoted 8 lines

arch/x86/kvm/x86.c static int kernel_pio(struct kvm_vcpu *vcpu, void *pd)

Quoted 8 lines

arch/x86/kvm/x86.c +1 -0

 	/* TODO: String I/O for in kernel device */
 	int r;
 
+	struct kvm_io_ext_data ext_data;
 
 	if (vcpu->arch.pio.in)
 
 		r = kvm_io_bus_read(vcpu->kvm, KVM_PIO_BUS, vcpu-

arch.pio.port,

Quoted 41 lines

Quoted 23 lines

Quoted 23 lines

arch/x86/kvm/x86.c +1 -1

@@ -3947,7 +3950,7 @@ static int kernel_pio(struct kvm_vcpu *vcpu,
 void *pd)
 
 	else
 
 		r = kvm_io_bus_write(vcpu->kvm, KVM_PIO_BUS,
 
 				     vcpu->arch.pio.port, vcpu->arch.pio.size,
 
-				     pd);
+				     pd, &ext_data);
 
 	return r;
 
 }

include/linux/kvm_host.h +0 -0

index 7d313e0..6bb211d 100644
@@ -68,8 +68,15 @@ enum kvm_bus {
 

Quoted 8 lines

include/linux/kvm_host.h enum kvm_bus { · line 72

Quoted 6 lines

include/linux/kvm_host.h +2 -0

 	KVM_NR_BUSES
 
 };
 
+struct kvm_io_ext_data {
+	int type;

What values does this get? Please add documentation in comments.

See the next patch.

Quoted 6 lines

include/linux/kvm_host.h enum kvm_bus { · line 75

Quoted 3 lines

include/linux/kvm_host.h +3 -0

+	union {
+		char padding[256];
+	};

So the structure size is 260 bytes?
What's the point of the padding?

Quoted 1 line

Reserved spaces. Also used in the next patch.

I was unable to find anything related to padding in the next patch. This is an internal API, isn't it? So why reserve space? Further, 256 bytes is quite a lot to reserve. Also,
making the total size a power of 2 would seem to make more sense if we do need to reserve any space.

Quoted 151 lines

--
regards
Yang, Sheng

Quoted 146 lines

Quoted 146 lines

include/linux/kvm_host.h +5 -2

+};
+
 
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t
 addr,
 
-		     int len, const void *val);
+		     int len, const void *val, struct kvm_io_ext_data *data);
 
 int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t
 addr, int len,
 
 		    void *val);
 
 int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx,

@@ -113,7 +120,8 @@ struct kvm_io_device_ops {
 
 	int (*write)(struct kvm_io_device *this,
 
 		     gpa_t addr,
 		     int len,
 
-		     const void *val);
+		     const void *val,
+		     struct kvm_io_ext_data *data);

include/linux/kvm_host.h +0 -0

 	void (*destructor)(struct kvm_io_device *this);
 
 };

virt/kvm/coalesced_mmio.c +2 -1

index fc84875..37b254c 100644
@@ -59,7 +59,8 @@ static int coalesced_mmio_in_range(struct
 kvm_coalesced_mmio_dev *dev,
 
 }
 
 static int coalesced_mmio_write(struct kvm_io_device *this,
 
-				gpa_t addr, int len, const void *val)
+				gpa_t addr, int len, const void *val,
+				struct kvm_io_ext_data *ext_data)

{

virt/kvm/coalesced_mmio.c +0 -0

 	struct kvm_coalesced_mmio_dev *dev = to_mmio(this);
 	struct kvm_coalesced_mmio_ring *ring =
 	dev->kvm->coalesced_mmio_ring;

virt/kvm/eventfd.c +1 -1

index 2ca4535..8edd757 100644
@@ -483,7 +483,7 @@ ioeventfd_in_range(struct _ioeventfd *p, gpa_t
 addr, int len, const void *val)
 
 /* MMIO/PIO writes trigger an event if the addr/val match */
 static int
 ioeventfd_write(struct kvm_io_device *this, gpa_t addr, int len,
 
-		const void *val)
+		const void *val, struct kvm_io_ext_data *ext_data)

{

virt/kvm/eventfd.c +0 -0

 	struct _ioeventfd *p = to_ioeventfd(this);

virt/kvm/ioapic.c +1 -1

index 0b9df83..6a027ef 100644
@@ -321,7 +321,7 @@ static int ioapic_mmio_read(struct kvm_io_device
 *this, gpa_t addr, int len,
 
 }
 
 static int ioapic_mmio_write(struct kvm_io_device *this, gpa_t addr,
 int len,
 
-			     const void *val)
+			     const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
 	struct kvm_ioapic *ioapic = to_ioapic(this);
 	u32 data;

virt/kvm/iodev.h +3 -3

index d1f5651..340ab79 100644
@@ -33,9 +33,11 @@ static inline int kvm_iodevice_read(struct
 kvm_io_device *dev,
 
 }
 
 static inline int kvm_iodevice_write(struct kvm_io_device *dev,
 
-				     gpa_t addr, int l, const void *v)
+				     gpa_t addr, int l, const void *v,
+				     struct kvm_io_ext_data *data)
 
 {
 
-	return dev->ops->write ? dev->ops->write(dev, addr, l, v) :
-EOPNOTSUPP; +	return dev->ops->write ?
+		dev->ops->write(dev, addr, l, v, data) : -EOPNOTSUPP;
 
 }
 
 static inline void kvm_iodevice_destructor(struct kvm_io_device
 *dev)

virt/kvm/kvm_main.c +1 -1

index b1b6cbb..a61f90e 100644
@@ -2221,14 +2221,14 @@ static void kvm_io_bus_destroy(struct
 kvm_io_bus *bus)
 
 /* kvm_io_bus_write - called under kvm->slots_lock */
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t
 addr,
 
-		     int len, const void *val)
+		     int len, const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
 	int i;
 	struct kvm_io_bus *bus;
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);

virt/kvm/kvm_main.c +1 -1

 	for (i = 0; i < bus->dev_count; i++)
 
-		if (!kvm_iodevice_write(bus->devs[i], addr, len, val))
+		if (!kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data))
 
 			return 0;
 
 	return -EOPNOTSUPP;
 
 }

Then we can support mask bit operation of assigned devices now.

Signed-off-by Sheng Yang <[email protected]>

---
 arch/x86/include/asm/kvm_host.h |    1 +
 arch/x86/kvm/Makefile           |    2 +-
 arch/x86/kvm/mmu.c              |    2 +
 arch/x86/kvm/x86.c              |   40 ++++-
 include/linux/kvm.h             |   28 ++++
 include/linux/kvm_host.h        |   34 +++++
 virt/kvm/assigned-dev.c         |   44 ++++++
 virt/kvm/kvm_main.c             |   38 +++++-
 virt/kvm/msix_mmio.c            |  296 +++++++++++++++++++++++++++++++++++++++
 virt/kvm/msix_mmio.h            |   25 ++++
 10 files changed, 497 insertions(+), 13 deletions(-)
 create mode 100644 virt/kvm/msix_mmio.c
 create mode 100644 virt/kvm/msix_mmio.h

arch/x86/include/asm/kvm_host.h +1 -0

index aa75f21..4a390a4 100644
@@ -635,6 +635,7 @@ enum emulation_result {
 	EMULATE_DONE,       /* no further processing */
 	EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
 	EMULATE_FAIL,         /* can't emulate this instruction */
+	EMULATE_USERSPACE_EXIT, /* we need exit to userspace */
 };
 
 #define EMULTYPE_NO_DECODE	    (1 << 0)

arch/x86/kvm/Makefile +1 -1

index f15501f..3a0d851 100644
@@ -7,7 +7,7 @@ CFLAGS_vmx.o := -I.
 
 kvm-y			+= $(addprefix ../../../virt/kvm/, kvm_main.o ioapic.o \
 				coalesced_mmio.o irq_comm.o eventfd.o \
-				assigned-dev.o)
+				assigned-dev.o msix_mmio.o)
 kvm-$(CONFIG_IOMMU_API)	+= $(addprefix ../../../virt/kvm/, iommu.o)
 kvm-$(CONFIG_KVM_ASYNC_PF)	+= $(addprefix ../../../virt/kvm/, async_pf.o)
 

arch/x86/kvm/mmu.c +2 -0

index 9cafbb4..912dca4 100644
@@ -3358,6 +3358,8 @@ int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t cr2, u32 error_code,
 	case EMULATE_DO_MMIO:
 		++vcpu->stat.mmio_exits;
 		/* fall through */
+	case EMULATE_USERSPACE_EXIT:
+		/* fall through */
 	case EMULATE_FAIL:
 		return 0;
 	default:

arch/x86/kvm/x86.c +32 -8

index 21b84e2..87308eb 100644
@@ -1966,6 +1966,7 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_X86_ROBUST_SINGLESTEP:
 	case KVM_CAP_XSAVE:
 	case KVM_CAP_ASYNC_PF:
+	case KVM_CAP_MSIX_MMIO:
 		r = 1;
 		break;
 	case KVM_CAP_COALESCED_MMIO:

@@ -3809,6 +3810,7 @@ static int emulator_write_emulated_onepage(unsigned long addr,
 {
 	gpa_t                 gpa;
 	struct kvm_io_ext_data ext_data;
+	int r;
 
 	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, exception);
 

@@ -3824,18 +3826,32 @@ static int emulator_write_emulated_onepage(unsigned long addr,
 
 mmio:
 	trace_kvm_mmio(KVM_TRACE_MMIO_WRITE, bytes, gpa, *(u64 *)val);
+	r = vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data);
 	/*
 	 * Is this MMIO handled locally?
 	 */
-	if (!vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data))
+	if (!r)
 		return X86EMUL_CONTINUE;
 
-	vcpu->mmio_needed = 1;
-	vcpu->run->exit_reason = KVM_EXIT_MMIO;
-	vcpu->run->mmio.phys_addr = vcpu->mmio_phys_addr = gpa;
-	vcpu->run->mmio.len = vcpu->mmio_size = bytes;
-	vcpu->run->mmio.is_write = vcpu->mmio_is_write = 1;
-	memcpy(vcpu->run->mmio.data, val, bytes);
+	if (r == -ENOTSYNC) {
+		vcpu->userspace_exit_needed = 1;
+		vcpu->run->exit_reason = KVM_EXIT_MSIX_ROUTING_UPDATE;
+		vcpu->run->msix_routing.dev_id =
+			ext_data.msix_routing.dev_id;
+		vcpu->run->msix_routing.type =
+			ext_data.msix_routing.type;
+		vcpu->run->msix_routing.entry_idx =
+			ext_data.msix_routing.entry_idx;
+		vcpu->run->msix_routing.flags =
+			ext_data.msix_routing.flags;
+	} else  {
+		vcpu->mmio_needed = 1;
+		vcpu->run->exit_reason = KVM_EXIT_MMIO;
+		vcpu->run->mmio.phys_addr = vcpu->mmio_phys_addr = gpa;
+		vcpu->run->mmio.len = vcpu->mmio_size = bytes;
+		vcpu->run->mmio.is_write = vcpu->mmio_is_write = 1;
+		memcpy(vcpu->run->mmio.data, val, bytes);
+	}
 
 	return X86EMUL_CONTINUE;
 }

@@ -4469,6 +4485,8 @@ done:
 		r = EMULATE_DO_MMIO;
 	} else if (r == EMULATION_RESTART)
 		goto restart;
+	else if (vcpu->userspace_exit_needed)
+		r = EMULATE_USERSPACE_EXIT;
 	else
 		r = EMULATE_DONE;
 

@@ -5397,12 +5415,18 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 		}
 	}
 
-	if (vcpu->arch.pio.count || vcpu->mmio_needed) {
+	if (vcpu->arch.pio.count || vcpu->mmio_needed ||
+			vcpu->userspace_exit_needed) {
 		if (vcpu->mmio_needed) {
 			memcpy(vcpu->mmio_data, kvm_run->mmio.data, 8);
 			vcpu->mmio_read_completed = 1;
 			vcpu->mmio_needed = 0;
 		}
+		if (vcpu->userspace_exit_needed) {
+			vcpu->userspace_exit_needed = 0;
+			r = 0;
+			goto out;
+		}
 		vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
 		r = emulate_instruction(vcpu, EMULTYPE_NO_DECODE);
 		srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);

include/linux/kvm.h +28 -0

index ea2dc1a..4393e4e 100644
@@ -161,6 +161,7 @@ struct kvm_pit_config {
 #define KVM_EXIT_NMI              16
 #define KVM_EXIT_INTERNAL_ERROR   17
 #define KVM_EXIT_OSI              18
+#define KVM_EXIT_MSIX_ROUTING_UPDATE 19
 
 /* For KVM_EXIT_INTERNAL_ERROR */
 #define KVM_INTERNAL_ERROR_EMULATION 1

@@ -264,6 +265,13 @@ struct kvm_run {
 		struct {
 			__u64 gprs[32];
 		} osi;
+		/* KVM_EXIT_MSIX_ROUTING_UPDATE*/
+		struct {
+			__u32 dev_id;
+			__u16 type;
+			__u16 entry_idx;
+			__u64 flags;
+		} msix_routing;
 		/* Fix the size of the union. */
 		char padding[256];
 	};

@@ -541,6 +549,7 @@ struct kvm_ppc_pvinfo {
 #define KVM_CAP_PPC_GET_PVINFO 57
 #define KVM_CAP_PPC_IRQ_LEVEL 58
 #define KVM_CAP_ASYNC_PF 59
+#define KVM_CAP_MSIX_MMIO 60
 
 #ifdef KVM_CAP_IRQ_ROUTING
 

@@ -672,6 +681,9 @@ struct kvm_clock_data {
 #define KVM_XEN_HVM_CONFIG        _IOW(KVMIO,  0x7a, struct kvm_xen_hvm_config)
 #define KVM_SET_CLOCK             _IOW(KVMIO,  0x7b, struct kvm_clock_data)
 #define KVM_GET_CLOCK             _IOR(KVMIO,  0x7c, struct kvm_clock_data)
+/* Available with KVM_CAP_MSIX_MMIO */
+#define KVM_REGISTER_MSIX_MMIO    _IOW(KVMIO,  0x7d, struct kvm_msix_mmio_user)
+#define KVM_UNREGISTER_MSIX_MMIO  _IOW(KVMIO,  0x7e, struct kvm_msix_mmio_user)
 /* Available with KVM_CAP_PIT_STATE2 */
 #define KVM_GET_PIT2              _IOR(KVMIO,  0x9f, struct kvm_pit_state2)
 #define KVM_SET_PIT2              _IOW(KVMIO,  0xa0, struct kvm_pit_state2)

@@ -795,4 +807,20 @@ struct kvm_assigned_msix_entry {
 	__u16 padding[3];
 };
 
+#define KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV	    (1 << 0)
+
+#define KVM_MSIX_MMIO_TYPE_BASE_TABLE	    (1 << 8)
+
+#define KVM_MSIX_MMIO_TYPE_DEV_MASK	    0x00ff
+#define KVM_MSIX_MMIO_TYPE_BASE_MASK	    0xff00
+struct kvm_msix_mmio_user {
+	__u32 dev_id;
+	__u16 type;
+	__u16 max_entries_nr;
+	__u64 base_addr;
+	__u64 base_va;
+	__u64 flags;
+	__u64 reserved[4];
+};
+
 #endif /* __LINUX_KVM_H */

include/linux/kvm_host.h +34 -0

index 6bb211d..6aaf85e 100644
@@ -68,9 +68,16 @@ enum kvm_bus {
 	KVM_NR_BUSES
 };
 
+#define KVM_IO_EXT_DATA_TYPE_MSIX_ROUTING   1
 struct kvm_io_ext_data {
 	int type;
 	union {
+		struct {
+			u32 dev_id;
+			u16 type;
+			u16 entry_idx;
+			u64 flags;
+		} msix_routing;
 		char padding[256];
 	};
 };

@@ -168,6 +175,8 @@ struct kvm_vcpu {
 	} async_pf;
 #endif
 
+	int userspace_exit_needed;
+
 	struct kvm_vcpu_arch arch;
 };
 

@@ -241,6 +250,27 @@ struct kvm_memslots {
 					KVM_PRIVATE_MEM_SLOTS];
 };
 
+#define KVM_MSIX_MMIO_MAX    32
+
+struct kvm_msix_mmio {
+	u32 dev_id;
+	u16 type;
+	u16 max_entries_nr;
+	u64 flags;
+	gpa_t table_base_addr;
+	hva_t table_base_va;
+	gpa_t pba_base_addr;
+	hva_t pba_base_va;
+};
+
+struct kvm_msix_mmio_dev {
+	struct kvm *kvm;
+	struct kvm_io_device table_dev;
+	int mmio_nr;
+	struct kvm_msix_mmio mmio[KVM_MSIX_MMIO_MAX];
+	struct mutex lock;
+};
+
 struct kvm {
 	spinlock_t mmu_lock;
 	raw_spinlock_t requests_lock;

@@ -289,6 +319,7 @@ struct kvm {
 	long mmu_notifier_count;
 #endif
 	long tlbs_dirty;
+	struct kvm_msix_mmio_dev msix_mmio_dev;
 };
 
 /* The guest did something we don't support. */

@@ -561,6 +592,9 @@ void kvm_unregister_irq_ack_notifier(struct kvm *kvm,
 int kvm_request_irq_source_id(struct kvm *kvm);
 void kvm_free_irq_source_id(struct kvm *kvm, int irq_source_id);
 
+int kvm_assigned_device_update_msix_mask_bit(struct kvm *kvm,
+			int assigned_dev_id, int entry, bool mask);
+
 /* For vcpu->arch.iommu_flags */
 #define KVM_IOMMU_CACHE_COHERENCY	0x1
 

virt/kvm/assigned-dev.c +44 -0

index ae72ae6..d1598a6 100644
@@ -18,6 +18,7 @@
 #include <linux/interrupt.h>
 #include <linux/slab.h>
 #include "irq.h"
+#include "msix_mmio.h"
 
 static struct kvm_assigned_dev_kernel *kvm_find_assigned_dev(struct list_head *head,
 						      int assigned_dev_id)

@@ -191,12 +192,25 @@ static void kvm_free_assigned_irq(struct kvm *kvm,
 	kvm_deassign_irq(kvm, assigned_dev, assigned_dev->irq_requested_type);
 }
 
+static void assigned_device_free_msix_mmio(struct kvm *kvm,
+				struct kvm_assigned_dev_kernel *adev)
+{
+	struct kvm_msix_mmio mmio;
+
+	mmio.dev_id = adev->assigned_dev_id;
+	mmio.type = KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV |
+		    KVM_MSIX_MMIO_TYPE_BASE_TABLE;
+	kvm_free_msix_mmio(kvm, &mmio);
+}
+
 static void kvm_free_assigned_device(struct kvm *kvm,
 				     struct kvm_assigned_dev_kernel
 				     *assigned_dev)
 {
 	kvm_free_assigned_irq(kvm, assigned_dev);
 
+	assigned_device_free_msix_mmio(kvm, assigned_dev);
+
 	__pci_reset_function(assigned_dev->dev);
 	pci_restore_state(assigned_dev->dev);
 

@@ -785,3 +799,33 @@ out:
 	return r;
 }
 
+/* The caller should hold kvm->lock */
+int kvm_assigned_device_update_msix_mask_bit(struct kvm *kvm,
+				int assigned_dev_id, int entry, bool mask)
+{
+	int r = -EFAULT;
+	struct kvm_assigned_dev_kernel *adev;
+	int i;
+
+	if (!irqchip_in_kernel(kvm))
+		return r;
+
+	adev = kvm_find_assigned_dev(&kvm->arch.assigned_dev_head,
+				      assigned_dev_id);
+	if (!adev)
+		goto out;
+
+	/* For non-MSIX enabled devices, entries_nr == 0 */
+	for (i = 0; i < adev->entries_nr; i++)
+		if (adev->host_msix_entries[i].entry == entry) {
+			if (mask)
+				disable_irq_nosync(
+					adev->host_msix_entries[i].vector);
+			else
+				enable_irq(adev->host_msix_entries[i].vector);
+			r = 0;
+			break;
+		}
+out:
+	return r;
+}

virt/kvm/kvm_main.c +34 -4

index a61f90e..f211e49 100644
@@ -56,6 +56,7 @@
 
 #include "coalesced_mmio.h"
 #include "async_pf.h"
+#include "msix_mmio.h"
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/kvm.h>

@@ -509,6 +510,7 @@ static void kvm_destroy_vm(struct kvm *kvm)
 	struct mm_struct *mm = kvm->mm;
 
 	kvm_arch_sync_events(kvm);
+	kvm_unregister_msix_mmio_dev(kvm);
 	spin_lock(&kvm_lock);
 	list_del(&kvm->vm_list);
 	spin_unlock(&kvm_lock);

@@ -1877,6 +1879,24 @@ static long kvm_vm_ioctl(struct file *filp,
 		mutex_unlock(&kvm->lock);
 		break;
 #endif
+	case KVM_REGISTER_MSIX_MMIO: {
+		struct kvm_msix_mmio_user mmio_user;
+
+		r = -EFAULT;
+		if (copy_from_user(&mmio_user, argp, sizeof mmio_user))
+			goto out;
+		r = kvm_vm_ioctl_register_msix_mmio(kvm, &mmio_user);
+		break;
+	}
+	case KVM_UNREGISTER_MSIX_MMIO: {
+		struct kvm_msix_mmio_user mmio_user;
+
+		r = -EFAULT;
+		if (copy_from_user(&mmio_user, argp, sizeof mmio_user))
+			goto out;
+		r = kvm_vm_ioctl_unregister_msix_mmio(kvm, &mmio_user);
+		break;
+	}
 	default:
 		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
 		if (r == -ENOTTY)

@@ -1988,6 +2008,12 @@ static int kvm_dev_ioctl_create_vm(void)
 		return r;
 	}
 #endif
+	r = kvm_register_msix_mmio_dev(kvm);
+	if (r < 0) {
+		kvm_put_kvm(kvm);
+		return r;
+	}
+
 	r = anon_inode_getfd("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
 	if (r < 0)
 		kvm_put_kvm(kvm);

@@ -2223,14 +2249,18 @@ static void kvm_io_bus_destroy(struct kvm_io_bus *bus)
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 		     int len, const void *val, struct kvm_io_ext_data *ext_data)
 {
-	int i;
+	int i, r = -EOPNOTSUPP;
 	struct kvm_io_bus *bus;
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
-	for (i = 0; i < bus->dev_count; i++)
-		if (!kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data))
+	for (i = 0; i < bus->dev_count; i++) {
+		r = kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data);
+		if (r == -ENOTSYNC)
+			break;
+		else if (!r)
 			return 0;
-	return -EOPNOTSUPP;
+	}
+	return r;
 }
 
 /* kvm_io_bus_read - called under kvm->slots_lock */

virt/kvm/msix_mmio.c new file +54 -0

new file mode 100644
index 0000000..083b15b
@@ -0,0 +1,296 @@
+/*
+ * MSI-X MMIO emulation
+ *
+ * Copyright (c) 2010 Intel Corporation
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ * Author:
+ *   Sheng Yang <[email protected]>
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/kvm.h>
+
+#include "msix_mmio.h"
+#include "iodev.h"
+
+static int update_msix_mask_bit(struct kvm *kvm, struct kvm_msix_mmio *mmio,
+				int entry, u32 flag)
+{
+	if (mmio->type & KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV)
+		return kvm_assigned_device_update_msix_mask_bit(kvm,
+				mmio->dev_id, entry, flag);
+	return -EFAULT;
+}
+
+/* Caller must hold dev->lock */
+static int get_mmio_table_index(struct kvm_msix_mmio_dev *dev,
+				gpa_t addr, int len)
+{
+	gpa_t start, end;
+	int i, r = -EINVAL;
+
+	for (i = 0; i < dev->mmio_nr; i++) {
+		start = dev->mmio[i].table_base_addr;
+		end = dev->mmio[i].table_base_addr + PCI_MSIX_ENTRY_SIZE *
+			dev->mmio[i].max_entries_nr;
+		if (addr >= start && addr + len <= end) {
+			r = i;
+			break;
+		}
+	}
+
+	return r;
+}
+
+static int msix_table_mmio_read(struct kvm_io_device *this, gpa_t addr, int len,
+				void *val)
+{
+	/*TODO: Add big endian support */
+	struct kvm_msix_mmio_dev *mmio_dev =
+		container_of(this, struct kvm_msix_mmio_dev, table_dev);
+	struct kvm_msix_mmio *mmio;

virt/kvm/msix_mmio.c new file +14 -0

+
+	mmio = &mmio_dev->mmio[idx];
+	entry = (addr - mmio->table_base_addr) / PCI_MSIX_ENTRY_SIZE;
+	r = copy_from_user(val, (void __user *)(mmio->table_base_va +
+			entry * PCI_MSIX_ENTRY_SIZE + offset), len);
+	if (r)
+		goto out;
+out:
+	mutex_unlock(&mmio_dev->lock);
+	return ret;
+}
+
+static int msix_table_mmio_write(struct kvm_io_device *this, gpa_t addr,
+				int len, const void *val,

virt/kvm/msix_mmio.c new file +107 -0

+
+	if (get_user(old_ctrl, ctrl_pos))
+		goto out;
+
+	/* Don't allow writing to other fields when entry is unmasked */
+	if (!(old_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT) &&
+	    offset != PCI_MSIX_ENTRY_VECTOR_CTRL)
+		goto out;
+
+	if (copy_to_user((void __user *)(entry_base + offset), val, len))
+		goto out;
+
+	ext_data->type = KVM_IO_EXT_DATA_TYPE_MSIX_ROUTING;
+	ext_data->msix_routing.dev_id = mmio->dev_id;
+	ext_data->msix_routing.type = mmio->type;
+	ext_data->msix_routing.entry_idx = entry;
+	ext_data->msix_routing.flags = 0;
+
+	if (offset + len < PCI_MSIX_ENTRY_VECTOR_CTRL) {
+		ret = -ENOTSYNC;
+		goto out;
+	}
+
+	if (get_user(new_ctrl, ctrl_pos))
+		goto out;
+
+	if (old_ctrl == new_ctrl) {
+		if (offset == PCI_MSIX_ENTRY_DATA && len == 8)
+			ret = -ENOTSYNC;
+		goto out;
+	}
+	if ((old_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT) ^
+			(new_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT))
+		r = update_msix_mask_bit(mmio_dev->kvm, mmio, entry,
+				!!(new_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT));
+	if (r)
+		ret = -ENOTSYNC;
+out:
+	mutex_unlock(&mmio_dev->lock);
+	mutex_unlock(&mmio_dev->kvm->lock);
+	return ret;
+}
+
+static const struct kvm_io_device_ops msix_mmio_table_ops = {
+	.read     = msix_table_mmio_read,
+	.write    = msix_table_mmio_write,
+};
+
+int kvm_register_msix_mmio_dev(struct kvm *kvm)
+{
+	int ret;
+
+	kvm_iodevice_init(&kvm->msix_mmio_dev.table_dev, &msix_mmio_table_ops);
+	mutex_init(&kvm->msix_mmio_dev.lock);
+	kvm->msix_mmio_dev.kvm = kvm;
+	mutex_lock(&kvm->slots_lock);
+	ret = kvm_io_bus_register_dev(kvm, KVM_MMIO_BUS,
+				      &kvm->msix_mmio_dev.table_dev);
+	mutex_unlock(&kvm->slots_lock);
+	return ret;
+}
+
+int kvm_unregister_msix_mmio_dev(struct kvm *kvm)
+{
+	int ret;
+
+	mutex_lock(&kvm->slots_lock);
+	ret = kvm_io_bus_unregister_dev(kvm, KVM_MMIO_BUS,
+				      &kvm->msix_mmio_dev.table_dev);
+	mutex_unlock(&kvm->slots_lock);
+	return ret;
+}
+
+int kvm_vm_ioctl_register_msix_mmio(struct kvm *kvm,
+				    struct kvm_msix_mmio_user *mmio_user)
+{
+	struct kvm_msix_mmio_dev *mmio_dev = &kvm->msix_mmio_dev;
+	struct kvm_msix_mmio *mmio = NULL;
+	int r = 0, i;
+
+	mutex_lock(&mmio_dev->lock);
+	for (i = 0; i < mmio_dev->mmio_nr; i++) {
+		if (mmio_dev->mmio[i].dev_id == mmio_user->dev_id &&
+		    (mmio_dev->mmio[i].type & KVM_MSIX_MMIO_TYPE_DEV_MASK) ==
+		    (mmio_user->type & KVM_MSIX_MMIO_TYPE_DEV_MASK)) {
+			mmio = &mmio_dev->mmio[i];
+			if (mmio->max_entries_nr != mmio_user->max_entries_nr) {
+				r = -EINVAL;
+				goto out;
+			}
+			break;
+		}
+	}
+	if (mmio_user->max_entries_nr > KVM_MAX_MSIX_PER_DEV) {
+		r = -EINVAL;
+		goto out;
+	}
+	/* All reserved currently */
+	if (mmio_user->flags) {
+		r = -EINVAL;
+		goto out;
+	}
+
+	if ((mmio_user->type & KVM_MSIX_MMIO_TYPE_DEV_MASK) !=
+			KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV) {
+		r = -EINVAL;
+		goto out;

virt/kvm/msix_mmio.c new file +69 -0

+			mmio_user->max_entries_nr * PCI_MSIX_ENTRY_SIZE)) {
+		r = -EINVAL;
+		goto out;
+	}
+	if (!mmio) {
+		if (mmio_dev->mmio_nr == KVM_MSIX_MMIO_MAX) {
+			r = -ENOSPC;
+			goto out;
+		}
+		mmio = &mmio_dev->mmio[mmio_dev->mmio_nr];
+		mmio_dev->mmio_nr++;
+	}
+
+	mmio->max_entries_nr = mmio_user->max_entries_nr;
+	mmio->dev_id = mmio_user->dev_id;
+	mmio->flags = mmio_user->flags;
+
+	if ((mmio_user->type & KVM_MSIX_MMIO_TYPE_DEV_MASK) ==
+			KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV)
+		mmio->type = KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV;
+	if ((mmio_user->type & KVM_MSIX_MMIO_TYPE_BASE_MASK) ==
+			KVM_MSIX_MMIO_TYPE_BASE_TABLE) {
+		mmio->type |= KVM_MSIX_MMIO_TYPE_BASE_TABLE;
+		mmio->table_base_addr = mmio_user->base_addr;
+		mmio->table_base_va = mmio_user->base_va;
+	}
+out:
+	mutex_unlock(&mmio_dev->lock);
+	return r;
+}
+
+int kvm_free_msix_mmio(struct kvm *kvm, struct kvm_msix_mmio *mmio)
+{
+	struct kvm_msix_mmio_dev *mmio_dev = &kvm->msix_mmio_dev;
+	int r = -EINVAL, i, j;
+
+	if (!mmio)
+		return 0;
+
+	mutex_lock(&mmio_dev->lock);
+	BUG_ON(mmio_dev->mmio_nr > KVM_MSIX_MMIO_MAX);
+	for (i = 0; i < mmio_dev->mmio_nr; i++) {
+		if (mmio_dev->mmio[i].dev_id == mmio->dev_id &&
+		    mmio_dev->mmio[i].type == mmio->type) {
+			r = 0;
+			for (j = i; j < mmio_dev->mmio_nr - 1; j++)
+				mmio_dev->mmio[j] = mmio_dev->mmio[j + 1];
+			mmio_dev->mmio[mmio_dev->mmio_nr].max_entries_nr = 0;
+			mmio_dev->mmio[mmio_dev->mmio_nr].dev_id = 0;
+			mmio_dev->mmio[mmio_dev->mmio_nr].type = 0;
+			mmio_dev->mmio_nr--;
+			break;
+		}
+	}
+	mutex_unlock(&mmio_dev->lock);
+	return r;
+}
+
+int kvm_vm_ioctl_unregister_msix_mmio(struct kvm *kvm,
+				      struct kvm_msix_mmio_user *mmio_user)
+{
+	struct kvm_msix_mmio mmio;
+
+	mmio.dev_id = mmio_user->dev_id;
+	mmio.type = mmio_user->type;
+
+	return kvm_free_msix_mmio(kvm, &mmio);
+}
+

virt/kvm/msix_mmio.h new file +25 -0

new file mode 100644
index 0000000..01b6587
@@ -0,0 +1,25 @@
+#ifndef __KVM_MSIX_MMIO_H__
+#define __KVM_MSIX_MMIO_H__
+/*
+ * MSI-X MMIO emulation
+ *
+ * Copyright (c) 2010 Intel Corporation
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ * Author:
+ *   Sheng Yang <[email protected]>
+ */
+
+#include <linux/pci.h>
+
+int kvm_register_msix_mmio_dev(struct kvm *kvm);
+int kvm_unregister_msix_mmio_dev(struct kvm *kvm);
+int kvm_vm_ioctl_register_msix_mmio(struct kvm *kvm,
+				    struct kvm_msix_mmio_user *mmio_user);
+int kvm_vm_ioctl_unregister_msix_mmio(struct kvm *kvm,
+				      struct kvm_msix_mmio_user *mmio_user);
+int kvm_free_msix_mmio(struct kvm *kvm, struct kvm_msix_mmio *mmio_user);
+
+#endif

Signature · 1 line

1.7.0.1

Quoted 3 lines · Sheng Yang

Then we can support mask bit operation of assigned devices now.

Signed-off-by Sheng Yang <[email protected]>

Doesn't look like all comments got addressed. E.g. gpa_t entry_base is still there and in reality you said it's a host virtual address so should be void __user *;

And ENOTSYNC meaning 'MSIX' is pretty hacky.

Quoted 632 lines

---

 arch/x86/include/asm/kvm_host.h |    1 +
 arch/x86/kvm/Makefile           |    2 +-
 arch/x86/kvm/mmu.c              |    2 +
 arch/x86/kvm/x86.c              |   40 ++++-
 include/linux/kvm.h             |   28 ++++
 include/linux/kvm_host.h        |   34 +++++
 virt/kvm/assigned-dev.c         |   44 ++++++
 virt/kvm/kvm_main.c             |   38 +++++-
 virt/kvm/msix_mmio.c            |  296
 +++++++++++++++++++++++++++++++++++++++ virt/kvm/msix_mmio.h           
 |   25 ++++
 10 files changed, 497 insertions(+), 13 deletions(-)
 create mode 100644 virt/kvm/msix_mmio.c
 create mode 100644 virt/kvm/msix_mmio.h

(quoted diff) +0 -0

b/arch/x86/include/asm/kvm_host.h index aa75f21..4a390a4 100644

arch/x86/include/asm/kvm_host.h +1 -0

@@ -635,6 +635,7 @@ enum emulation_result {
 
 	EMULATE_DONE,       /* no further processing */
 	EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
 	EMULATE_FAIL,         /* can't emulate this instruction */
 
+	EMULATE_USERSPACE_EXIT, /* we need exit to userspace */
 

};

#define EMULTYPE_NO_DECODE (1 << 0)

arch/x86/kvm/Makefile +1 -1

index f15501f..3a0d851 100644
@@ -7,7 +7,7 @@ CFLAGS_vmx.o := -I.
 
 kvm-y			+= $(addprefix ../../../virt/kvm/, kvm_main.o ioapic.o \
 
 				coalesced_mmio.o irq_comm.o eventfd.o \
 
-				assigned-dev.o)
+				assigned-dev.o msix_mmio.o)
 

kvm-$(CONFIG_IOMMU_API) += $(addprefix ../../../virt/kvm/, iommu.o)
kvm-$(CONFIG_KVM_ASYNC_PF) += $(addprefix ../../../virt/kvm/,
async_pf.o)

arch/x86/kvm/mmu.c +2 -0

index 9cafbb4..912dca4 100644
@@ -3358,6 +3358,8 @@ int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t
 cr2, u32 error_code,
 
 	case EMULATE_DO_MMIO:
 		++vcpu->stat.mmio_exits;
 		/* fall through */
 
+	case EMULATE_USERSPACE_EXIT:
+		/* fall through */

arch/x86/kvm/mmu.c +0 -0

 	case EMULATE_FAIL:
 		return 0;
 
 	default:

arch/x86/kvm/x86.c +1 -0

index 21b84e2..87308eb 100644
@@ -1966,6 +1966,7 @@ int kvm_dev_ioctl_check_extension(long ext)
 
 	case KVM_CAP_X86_ROBUST_SINGLESTEP:
 	case KVM_CAP_XSAVE:
 
 	case KVM_CAP_ASYNC_PF:
+	case KVM_CAP_MSIX_MMIO:
 		r = 1;

arch/x86/kvm/x86.c +24 -7

 		break;
 
 	case KVM_CAP_COALESCED_MMIO:

@@ -3809,6 +3810,7 @@ static int emulator_write_emulated_onepage(unsigned
 long addr,
 
 {
 
 	gpa_t                 gpa;
 	struct kvm_io_ext_data ext_data;
 
+	int r;
 
 	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, exception);

@@ -3824,18 +3826,32 @@ static int
 emulator_write_emulated_onepage(unsigned long addr,
 
 mmio:
 	trace_kvm_mmio(KVM_TRACE_MMIO_WRITE, bytes, gpa, *(u64 *)val);
 
+	r = vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data);
 
 	/*
 
 	 * Is this MMIO handled locally?
 	 */
 
-	if (!vcpu_mmio_write(vcpu, gpa, bytes, val, &ext_data))
+	if (!r)
 
 		return X86EMUL_CONTINUE;
 
-	vcpu->mmio_needed = 1;
-	vcpu->run->exit_reason = KVM_EXIT_MMIO;
-	vcpu->run->mmio.phys_addr = vcpu->mmio_phys_addr = gpa;
-	vcpu->run->mmio.len = vcpu->mmio_size = bytes;
-	vcpu->run->mmio.is_write = vcpu->mmio_is_write = 1;
-	memcpy(vcpu->run->mmio.data, val, bytes);
+	if (r == -ENOTSYNC) {
+		vcpu->userspace_exit_needed = 1;
+		vcpu->run->exit_reason = KVM_EXIT_MSIX_ROUTING_UPDATE;
+		vcpu->run->msix_routing.dev_id =
+			ext_data.msix_routing.dev_id;
+		vcpu->run->msix_routing.type =
+			ext_data.msix_routing.type;
+		vcpu->run->msix_routing.entry_idx =
+			ext_data.msix_routing.entry_idx;
+		vcpu->run->msix_routing.flags =
+			ext_data.msix_routing.flags;
+	} else  {
+		vcpu->mmio_needed = 1;
+		vcpu->run->exit_reason = KVM_EXIT_MMIO;
+		vcpu->run->mmio.phys_addr = vcpu->mmio_phys_addr = gpa;
+		vcpu->run->mmio.len = vcpu->mmio_size = bytes;
+		vcpu->run->mmio.is_write = vcpu->mmio_is_write = 1;
+		memcpy(vcpu->run->mmio.data, val, bytes);
+	}
 
 	return X86EMUL_CONTINUE;
 
 }

@@ -4469,6 +4485,8 @@ done:
 		r = EMULATE_DO_MMIO;
 
 	} else if (r == EMULATION_RESTART)
 
 		goto restart;
 
+	else if (vcpu->userspace_exit_needed)
+		r = EMULATE_USERSPACE_EXIT;

arch/x86/kvm/x86.c +7 -1

 	else
 
 		r = EMULATE_DONE;

@@ -5397,12 +5415,18 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu
 *vcpu, struct kvm_run *kvm_run)
 
 		}
 
 	}
 
-	if (vcpu->arch.pio.count || vcpu->mmio_needed) {
+	if (vcpu->arch.pio.count || vcpu->mmio_needed ||
+			vcpu->userspace_exit_needed) {
 
 		if (vcpu->mmio_needed) {
 
 			memcpy(vcpu->mmio_data, kvm_run->mmio.data, 8);
 			vcpu->mmio_read_completed = 1;
 			vcpu->mmio_needed = 0;
 
 		}
 
+		if (vcpu->userspace_exit_needed) {
+			vcpu->userspace_exit_needed = 0;
+			r = 0;
+			goto out;
+		}
 
 		vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
 		r = emulate_instruction(vcpu, EMULTYPE_NO_DECODE);
 		srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);

include/linux/kvm.h +1 -0

index ea2dc1a..4393e4e 100644
@@ -161,6 +161,7 @@ struct kvm_pit_config {
 
 #define KVM_EXIT_NMI              16
 #define KVM_EXIT_INTERNAL_ERROR   17
 #define KVM_EXIT_OSI              18
 
+#define KVM_EXIT_MSIX_ROUTING_UPDATE 19
 

/* For KVM_EXIT_INTERNAL_ERROR */
#define KVM_INTERNAL_ERROR_EMULATION 1

include/linux/kvm.h +8 -0

@@ -264,6 +265,13 @@ struct kvm_run {
 
 		struct {
 
 			__u64 gprs[32];
 
 		} osi;
 
+		/* KVM_EXIT_MSIX_ROUTING_UPDATE*/
+		struct {
+			__u32 dev_id;
+			__u16 type;
+			__u16 entry_idx;
+			__u64 flags;
+		} msix_routing;
 
 		/* Fix the size of the union. */
 		char padding[256];
 
 	};

@@ -541,6 +549,7 @@ struct kvm_ppc_pvinfo {
 
 #define KVM_CAP_PPC_GET_PVINFO 57
 #define KVM_CAP_PPC_IRQ_LEVEL 58
 #define KVM_CAP_ASYNC_PF 59
 
+#define KVM_CAP_MSIX_MMIO 60
 

#ifdef KVM_CAP_IRQ_ROUTING

include/linux/kvm.h +2 -0

@@ -672,6 +681,9 @@ struct kvm_clock_data {
 
 #define KVM_XEN_HVM_CONFIG        _IOW(KVMIO,  0x7a, struct
 kvm_xen_hvm_config) #define KVM_SET_CLOCK             _IOW(KVMIO, 
 0x7b, struct kvm_clock_data) #define KVM_GET_CLOCK            
 _IOR(KVMIO,  0x7c, struct kvm_clock_data)
 
+/* Available with KVM_CAP_MSIX_MMIO */
+#define KVM_REGISTER_MSIX_MMIO    _IOW(KVMIO,  0x7d, struct

kvm_msix_mmio_user) +#define KVM_UNREGISTER_MSIX_MMIO _IOW(KVMIO, 0x7e, struct kvm_msix_mmio_user)

/* Available with KVM_CAP_PIT_STATE2 */
#define KVM_GET_PIT2 _IOR(KVMIO, 0x9f, struct
kvm_pit_state2) #define KVM_SET_PIT2 _IOW(KVMIO, 0xa0,
struct kvm_pit_state2)

include/linux/kvm.h +16 -0

@@ -795,4 +807,20 @@ struct kvm_assigned_msix_entry {
 
 	__u16 padding[3];
 
 };
 
+#define KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV	    (1 << 0)
+
+#define KVM_MSIX_MMIO_TYPE_BASE_TABLE	    (1 << 8)
+
+#define KVM_MSIX_MMIO_TYPE_DEV_MASK	    0x00ff
+#define KVM_MSIX_MMIO_TYPE_BASE_MASK	    0xff00
+struct kvm_msix_mmio_user {
+	__u32 dev_id;
+	__u16 type;
+	__u16 max_entries_nr;
+	__u64 base_addr;
+	__u64 base_va;
+	__u64 flags;
+	__u64 reserved[4];
+};
+
 
 #endif /* __LINUX_KVM_H */

include/linux/kvm_host.h +9 -0

index 6bb211d..6aaf85e 100644
@@ -68,9 +68,16 @@ enum kvm_bus {
 
 	KVM_NR_BUSES
 
 };
 
+#define KVM_IO_EXT_DATA_TYPE_MSIX_ROUTING   1
 
 struct kvm_io_ext_data {
 
 	int type;
 	union {
 
+		struct {
+			u32 dev_id;
+			u16 type;
+			u16 entry_idx;
+			u64 flags;
+		} msix_routing;
 
 		char padding[256];
 
 	};
 
 };

@@ -168,6 +175,8 @@ struct kvm_vcpu {
 
 	} async_pf;
 
 #endif
 
+	int userspace_exit_needed;
+
 

include/linux/kvm_host.h +21 -0

 	struct kvm_vcpu_arch arch;
 
 };

@@ -241,6 +250,27 @@ struct kvm_memslots {
 
 					KVM_PRIVATE_MEM_SLOTS];
 
 };
 
+#define KVM_MSIX_MMIO_MAX    32
+
+struct kvm_msix_mmio {
+	u32 dev_id;
+	u16 type;
+	u16 max_entries_nr;
+	u64 flags;
+	gpa_t table_base_addr;
+	hva_t table_base_va;
+	gpa_t pba_base_addr;
+	hva_t pba_base_va;
+};
+
+struct kvm_msix_mmio_dev {
+	struct kvm *kvm;
+	struct kvm_io_device table_dev;
+	int mmio_nr;
+	struct kvm_msix_mmio mmio[KVM_MSIX_MMIO_MAX];
+	struct mutex lock;
+};
+
 

struct kvm {

include/linux/kvm_host.h +4 -0

 	spinlock_t mmu_lock;
 	raw_spinlock_t requests_lock;

@@ -289,6 +319,7 @@ struct kvm {
 
 	long mmu_notifier_count;
 
 #endif
 
 	long tlbs_dirty;
 
+	struct kvm_msix_mmio_dev msix_mmio_dev;
 
 };
 
 /* The guest did something we don't support. */

@@ -561,6 +592,9 @@ void kvm_unregister_irq_ack_notifier(struct kvm *kvm,
 
 int kvm_request_irq_source_id(struct kvm *kvm);
 void kvm_free_irq_source_id(struct kvm *kvm, int irq_source_id);
 
+int kvm_assigned_device_update_msix_mask_bit(struct kvm *kvm,
+			int assigned_dev_id, int entry, bool mask);
+
 
 /* For vcpu->arch.iommu_flags */

#define KVM_IOMMU_CACHE_COHERENCY 0x1

virt/kvm/assigned-dev.c +1 -0

index ae72ae6..d1598a6 100644
@@ -18,6 +18,7 @@
 
 #include <linux/interrupt.h>
 #include <linux/slab.h>
 #include "irq.h"
 
+#include "msix_mmio.h"
 

static struct kvm_assigned_dev_kernel *kvm_find_assigned_dev(struct list_head *head,

virt/kvm/assigned-dev.c +43 -0

 						      int assigned_dev_id)

@@ -191,12 +192,25 @@ static void kvm_free_assigned_irq(struct kvm *kvm,
 
 	kvm_deassign_irq(kvm, assigned_dev, assigned_dev->irq_requested_type);
 
 }
 
+static void assigned_device_free_msix_mmio(struct kvm *kvm,
+				struct kvm_assigned_dev_kernel *adev)
+{
+	struct kvm_msix_mmio mmio;
+
+	mmio.dev_id = adev->assigned_dev_id;
+	mmio.type = KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV |
+		    KVM_MSIX_MMIO_TYPE_BASE_TABLE;
+	kvm_free_msix_mmio(kvm, &mmio);
+}
+
 
 static void kvm_free_assigned_device(struct kvm *kvm,
 
 				     struct kvm_assigned_dev_kernel
 				     *assigned_dev)
 
 {
 
 	kvm_free_assigned_irq(kvm, assigned_dev);
 
+	assigned_device_free_msix_mmio(kvm, assigned_dev);
+
 
 	__pci_reset_function(assigned_dev->dev);
 	pci_restore_state(assigned_dev->dev);

@@ -785,3 +799,33 @@ out:
 	return r;
 
 }
 
+/* The caller should hold kvm->lock */
+int kvm_assigned_device_update_msix_mask_bit(struct kvm *kvm,
+				int assigned_dev_id, int entry, bool mask)
+{
+	int r = -EFAULT;
+	struct kvm_assigned_dev_kernel *adev;
+	int i;
+
+	if (!irqchip_in_kernel(kvm))
+		return r;
+
+	adev = kvm_find_assigned_dev(&kvm->arch.assigned_dev_head,
+				      assigned_dev_id);
+	if (!adev)
+		goto out;
+
+	/* For non-MSIX enabled devices, entries_nr == 0 */
+	for (i = 0; i < adev->entries_nr; i++)
+		if (adev->host_msix_entries[i].entry == entry) {
+			if (mask)
+				disable_irq_nosync(
+					adev->host_msix_entries[i].vector);
+			else
+				enable_irq(adev->host_msix_entries[i].vector);
+			r = 0;
+			break;
+		}
+out:
+	return r;
+}

virt/kvm/kvm_main.c +1 -0

index a61f90e..f211e49 100644
@@ -56,6 +56,7 @@
 
 #include "coalesced_mmio.h"
 #include "async_pf.h"
 
+#include "msix_mmio.h"
 
 #define CREATE_TRACE_POINTS

#include <trace/events/kvm.h>

virt/kvm/kvm_main.c +1 -0

@@ -509,6 +510,7 @@ static void kvm_destroy_vm(struct kvm *kvm)
 
 	struct mm_struct *mm = kvm->mm;
 
 	kvm_arch_sync_events(kvm);
 
+	kvm_unregister_msix_mmio_dev(kvm);
 

virt/kvm/kvm_main.c +18 -0

 	spin_lock(&kvm_lock);
 	list_del(&kvm->vm_list);
 	spin_unlock(&kvm_lock);

@@ -1877,6 +1879,24 @@ static long kvm_vm_ioctl(struct file *filp,
 
 		mutex_unlock(&kvm->lock);
 		break;
 
 #endif
 
+	case KVM_REGISTER_MSIX_MMIO: {
+		struct kvm_msix_mmio_user mmio_user;
+
+		r = -EFAULT;
+		if (copy_from_user(&mmio_user, argp, sizeof mmio_user))
+			goto out;
+		r = kvm_vm_ioctl_register_msix_mmio(kvm, &mmio_user);
+		break;
+	}
+	case KVM_UNREGISTER_MSIX_MMIO: {
+		struct kvm_msix_mmio_user mmio_user;
+
+		r = -EFAULT;
+		if (copy_from_user(&mmio_user, argp, sizeof mmio_user))
+			goto out;
+		r = kvm_vm_ioctl_unregister_msix_mmio(kvm, &mmio_user);
+		break;
+	}

virt/kvm/kvm_main.c +14 -4

 	default:
 		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
 		if (r == -ENOTTY)

@@ -1988,6 +2008,12 @@ static int kvm_dev_ioctl_create_vm(void)
 
 		return r;
 
 	}
 
 #endif
 
+	r = kvm_register_msix_mmio_dev(kvm);
+	if (r < 0) {
+		kvm_put_kvm(kvm);
+		return r;
+	}
+
 
 	r = anon_inode_getfd("kvm-vm", &kvm_vm_fops, kvm, O_RDWR);
 	if (r < 0)
 
 		kvm_put_kvm(kvm);

@@ -2223,14 +2249,18 @@ static void kvm_io_bus_destroy(struct kvm_io_bus
 *bus)
 
 int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 
 		     int len, const void *val, struct kvm_io_ext_data *ext_data)
 
 {
 
-	int i;
+	int i, r = -EOPNOTSUPP;
 
 	struct kvm_io_bus *bus;
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
 
-	for (i = 0; i < bus->dev_count; i++)
-		if (!kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data))
+	for (i = 0; i < bus->dev_count; i++) {
+		r = kvm_iodevice_write(bus->devs[i], addr, len, val, ext_data);
+		if (r == -ENOTSYNC)
+			break;
+		else if (!r)
 
 			return 0;
 
-	return -EOPNOTSUPP;
+	}
+	return r;
 
 }
 
 /* kvm_io_bus_read - called under kvm->slots_lock */

virt/kvm/msix_mmio.c new file +19 -0

new file mode 100644
index 0000000..083b15b
@@ -0,0 +1,296 @@
+/*
+ * MSI-X MMIO emulation
+ *
+ * Copyright (c) 2010 Intel Corporation
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ * Author:
+ *   Sheng Yang <[email protected]>
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/kvm.h>
+
+#include "msix_mmio.h"
+#include "iodev.h"
+
+static int update_msix_mask_bit(struct kvm *kvm, struct kvm_msix_mmio

*mmio, + int entry, u32 flag)

virt/kvm/msix_mmio.c new file +28 -0

+{
+	if (mmio->type & KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV)
+		return kvm_assigned_device_update_msix_mask_bit(kvm,
+				mmio->dev_id, entry, flag);
+	return -EFAULT;
+}
+
+/* Caller must hold dev->lock */
+static int get_mmio_table_index(struct kvm_msix_mmio_dev *dev,
+				gpa_t addr, int len)
+{
+	gpa_t start, end;
+	int i, r = -EINVAL;
+
+	for (i = 0; i < dev->mmio_nr; i++) {
+		start = dev->mmio[i].table_base_addr;
+		end = dev->mmio[i].table_base_addr + PCI_MSIX_ENTRY_SIZE *
+			dev->mmio[i].max_entries_nr;
+		if (addr >= start && addr + len <= end) {
+			r = i;
+			break;
+		}
+	}
+
+	return r;
+}
+
+static int msix_table_mmio_read(struct kvm_io_device *this, gpa_t addr,

int len, + void *val)

virt/kvm/msix_mmio.c new file +5 -0

+{
+	/*TODO: Add big endian support */
+	struct kvm_msix_mmio_dev *mmio_dev =
+		container_of(this, struct kvm_msix_mmio_dev, table_dev);
+	struct kvm_msix_mmio *mmio;

virt/kvm/msix_mmio.c line 64

Quoted 10 lines

virt/kvm/msix_mmio.c new file +10 -0

+	int idx, ret = 0, entry, offset, r;
+
+	mutex_lock(&mmio_dev->lock);
+	idx = get_mmio_table_index(mmio_dev, addr, len);
+	if (idx < 0) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+	if ((addr & 0x3) || (len != 4 && len != 8))
+		goto out;

addr & len as below?

virt/kvm/msix_mmio.c line 68

Quoted 4 lines

virt/kvm/msix_mmio.c new file +4 -0

+
+	offset = addr % PCI_MSIX_ENTRY_SIZE;
+	if (offset == PCI_MSIX_ENTRY_VECTOR_CTRL && len == 8)
+		goto out;

then this test won't be needed.

Quoted 21 lines

virt/kvm/msix_mmio.c new file +21 -0

+
+	mmio = &mmio_dev->mmio[idx];
+	entry = (addr - mmio->table_base_addr) / PCI_MSIX_ENTRY_SIZE;
+	r = copy_from_user(val, (void __user *)(mmio->table_base_va +
+			entry * PCI_MSIX_ENTRY_SIZE + offset), len);
+	if (r)
+		goto out;
+out:
+	mutex_unlock(&mmio_dev->lock);
+	return ret;
+}
+
+static int msix_table_mmio_write(struct kvm_io_device *this, gpa_t addr,
+				int len, const void *val,
+				struct kvm_io_ext_data *ext_data)
+{
+	/*TODO: Add big endian support */
+	struct kvm_msix_mmio_dev *mmio_dev =
+		container_of(this, struct kvm_msix_mmio_dev, table_dev);
+	struct kvm_msix_mmio *mmio;
+	int idx, entry, offset, ret = 0, r = 0;

virt/kvm/msix_mmio.c line 92

Quoted 3 lines

virt/kvm/msix_mmio.c new file +3 -0

+	gpa_t entry_base;
+	u32 old_ctrl, new_ctrl;
+	unsigned long __user *ctrl_pos;

long? It's 8 bytes on 64 bit.
You really want
__le32 old_ctrl, new_ctrl;
__le32 __user *ctrl_pos;

virt/kvm/msix_mmio.c line 101

Quoted 10 lines

virt/kvm/msix_mmio.c new file +9 -0

+
+	mutex_lock(&mmio_dev->kvm->lock);
+	mutex_lock(&mmio_dev->lock);
+	idx = get_mmio_table_index(mmio_dev, addr, len);
+	if (idx < 0) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+	if (!(len == 4 || len == 8) || addr & (len - 1))

Nice hack. Even a bit nicer

	if ((len != 4 && len != 8) || addr & (len - 1))

virt/kvm/msix_mmio.c line 110

Quoted 10 lines

virt/kvm/msix_mmio.c new file +9 -0

+		goto out;
+
+	offset = addr % PCI_MSIX_ENTRY_SIZE;
+
+	mmio = &mmio_dev->mmio[idx];
+	entry = (addr - mmio->table_base_addr) / PCI_MSIX_ENTRY_SIZE;
+	entry_base = mmio->table_base_va + entry * PCI_MSIX_ENTRY_SIZE;
+	ctrl_pos = (unsigned long __user *)(entry_base +
+			PCI_MSIX_ENTRY_VECTOR_CTRL);

So this is the issue: if you cast a type to unsigned long * compiler can assume that the address is aligned. To prevent problems please add a check that table_base_va is aligned.

Quoted 113 lines

virt/kvm/msix_mmio.c new file +112 -0

+
+	if (get_user(old_ctrl, ctrl_pos))
+		goto out;
+
+	/* Don't allow writing to other fields when entry is unmasked */
+	if (!(old_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT) &&
+	    offset != PCI_MSIX_ENTRY_VECTOR_CTRL)
+		goto out;
+
+	if (copy_to_user((void __user *)(entry_base + offset), val, len))
+		goto out;
+
+	ext_data->type = KVM_IO_EXT_DATA_TYPE_MSIX_ROUTING;
+	ext_data->msix_routing.dev_id = mmio->dev_id;
+	ext_data->msix_routing.type = mmio->type;
+	ext_data->msix_routing.entry_idx = entry;
+	ext_data->msix_routing.flags = 0;
+
+	if (offset + len < PCI_MSIX_ENTRY_VECTOR_CTRL) {
+		ret = -ENOTSYNC;
+		goto out;
+	}
+
+	if (get_user(new_ctrl, ctrl_pos))
+		goto out;
+
+	if (old_ctrl == new_ctrl) {
+		if (offset == PCI_MSIX_ENTRY_DATA && len == 8)
+			ret = -ENOTSYNC;
+		goto out;
+	}
+	if ((old_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT) ^
+			(new_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT))
+		r = update_msix_mask_bit(mmio_dev->kvm, mmio, entry,
+				!!(new_ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT));
+	if (r)
+		ret = -ENOTSYNC;
+out:
+	mutex_unlock(&mmio_dev->lock);
+	mutex_unlock(&mmio_dev->kvm->lock);
+	return ret;
+}
+
+static const struct kvm_io_device_ops msix_mmio_table_ops = {
+	.read     = msix_table_mmio_read,
+	.write    = msix_table_mmio_write,
+};
+
+int kvm_register_msix_mmio_dev(struct kvm *kvm)
+{
+	int ret;
+
+	kvm_iodevice_init(&kvm->msix_mmio_dev.table_dev, &msix_mmio_table_ops);
+	mutex_init(&kvm->msix_mmio_dev.lock);
+	kvm->msix_mmio_dev.kvm = kvm;
+	mutex_lock(&kvm->slots_lock);
+	ret = kvm_io_bus_register_dev(kvm, KVM_MMIO_BUS,
+				      &kvm->msix_mmio_dev.table_dev);
+	mutex_unlock(&kvm->slots_lock);
+	return ret;
+}
+
+int kvm_unregister_msix_mmio_dev(struct kvm *kvm)
+{
+	int ret;
+
+	mutex_lock(&kvm->slots_lock);
+	ret = kvm_io_bus_unregister_dev(kvm, KVM_MMIO_BUS,
+				      &kvm->msix_mmio_dev.table_dev);
+	mutex_unlock(&kvm->slots_lock);
+	return ret;
+}
+
+int kvm_vm_ioctl_register_msix_mmio(struct kvm *kvm,
+				    struct kvm_msix_mmio_user *mmio_user)
+{
+	struct kvm_msix_mmio_dev *mmio_dev = &kvm->msix_mmio_dev;
+	struct kvm_msix_mmio *mmio = NULL;
+	int r = 0, i;
+
+	mutex_lock(&mmio_dev->lock);
+	for (i = 0; i < mmio_dev->mmio_nr; i++) {
+		if (mmio_dev->mmio[i].dev_id == mmio_user->dev_id &&
+		    (mmio_dev->mmio[i].type & KVM_MSIX_MMIO_TYPE_DEV_MASK) ==
+		    (mmio_user->type & KVM_MSIX_MMIO_TYPE_DEV_MASK)) {
+			mmio = &mmio_dev->mmio[i];
+			if (mmio->max_entries_nr != mmio_user->max_entries_nr) {
+				r = -EINVAL;
+				goto out;
+			}
+			break;
+		}
+	}
+	if (mmio_user->max_entries_nr > KVM_MAX_MSIX_PER_DEV) {
+		r = -EINVAL;
+		goto out;
+	}
+	/* All reserved currently */
+	if (mmio_user->flags) {
+		r = -EINVAL;
+		goto out;
+	}
+
+	if ((mmio_user->type & KVM_MSIX_MMIO_TYPE_DEV_MASK) !=
+			KVM_MSIX_MMIO_TYPE_ASSIGNED_DEV) {
+		r = -EINVAL;
+		goto out;
+	}
+	if ((mmio_user->type & KVM_MSIX_MMIO_TYPE_BASE_MASK) !=
+			KVM_MSIX_MMIO_TYPE_BASE_TABLE) {
+		r = -EINVAL;
+		goto out;