[Qemu-devel] [PATCH v2 0/5] KVM: irqfds for s390

[Qemu-devel] [PATCH v2 0/5] KVM: irqfds for s390

[Cornelia Huck]

Hi,

here's the next iteration of my patchset introducing irqfds for s390.

Changes from v1:
- rebased against current kvm/queue, bumping capability numbers
- guest adapter interrupt support is already in the kvm tree

I'd really like to get some feedback for the architecture independent
changes (and perhaps some thoughts on the restrictions described in the
last patch).

Again, find it on

git://git.kernel.org/pub/scm/linux/kernel/git/kvms390/linux.git kvms390-irqfd

Cornelia Huck (5):
  KVM: eventfd: Fix lock order inversion.
  KVM: Add per-vm capability enablement.
  KVM: s390: adapter interrupt sources
  KVM: s390: irq routing for adapter interrupts.
  KVM: Bump KVM_MAX_IRQ_ROUTES for s390

 Documentation/virtual/kvm/api.txt               |   27 ++-
 Documentation/virtual/kvm/devices/s390_flic.txt |   45 ++++
 arch/s390/include/asm/kvm_host.h                |   30 +++
 arch/s390/include/uapi/asm/kvm.h                |   22 ++
 arch/s390/kvm/Kconfig                           |    2 +
 arch/s390/kvm/Makefile                          |    2 +-
 arch/s390/kvm/interrupt.c                       |  287 ++++++++++++++++++++++-
 arch/s390/kvm/irq.h                             |   22 ++
 arch/s390/kvm/kvm-s390.c                        |   42 ++++
 arch/s390/kvm/kvm-s390.h                        |    2 +
 include/linux/kvm_host.h                        |   13 +
 include/uapi/linux/kvm.h                        |   16 ++
 virt/kvm/eventfd.c                              |    8 +-
 13 files changed, 506 insertions(+), 12 deletions(-)
 create mode 100644 arch/s390/kvm/irq.h

-- 
1.7.9.5

[Qemu-devel] [PATCH v2 1/5] KVM: eventfd: Fix lock order inversion.

[Cornelia Huck]

When registering a new irqfd, we call its ->poll method to collect any
event that might have previously been pending so that we can trigger it.
This is done under the kvm->irqfds.lock, which means the eventfd's ctx
lock is taken under it.

However, if we get a POLLHUP in irqfd_wakeup, we will be called with the
ctx lock held before getting the irqfds.lock to deactivate the irqfd,
causing lockdep to complain.

Calling the ->poll method does not really need the irqfds.lock, so let's
just move it after we've given up the irqfds.lock in kvm_irqfd_assign().

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
 virt/kvm/eventfd.c |    8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index abe4d60..29c2a04 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -391,19 +391,19 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
 					   lockdep_is_held(&kvm->irqfds.lock));
 	irqfd_update(kvm, irqfd, irq_rt);
 
-	events = f.file->f_op->poll(f.file, &irqfd->pt);
-
 	list_add_tail(&irqfd->list, &kvm->irqfds.items);
 
+	spin_unlock_irq(&kvm->irqfds.lock);
+
 	/*
 	 * Check if there was an event already pending on the eventfd
 	 * before we registered, and trigger it as if we didn't miss it.
 	 */
+	events = f.file->f_op->poll(f.file, &irqfd->pt);
+
 	if (events & POLLIN)
 		schedule_work(&irqfd->inject);
 
-	spin_unlock_irq(&kvm->irqfds.lock);
-
 	/*
 	 * do not drop the file until the irqfd is fully initialized, otherwise
 	 * we might race against the POLLHUP
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v2 1/5] KVM: eventfd: Fix lock order inversion.

[Christian Borntraeger]

On 17/03/14 19:11, Cornelia Huck wrote:
> When registering a new irqfd, we call its ->poll method to collect any
> event that might have previously been pending so that we can trigger it.
> This is done under the kvm->irqfds.lock, which means the eventfd's ctx
> lock is taken under it.
> 
> However, if we get a POLLHUP in irqfd_wakeup, we will be called with the
> ctx lock held before getting the irqfds.lock to deactivate the irqfd,
> causing lockdep to complain.
> 
> Calling the ->poll method does not really need the irqfds.lock, so let's
> just move it after we've given up the irqfds.lock in kvm_irqfd_assign().
> 
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Do you still have the lockdep message somewhere? 
Looking at the patch and the description this makes sense. Even without
irqfd for s390:
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>

Paolo, maybe this patch can go in independently from s390?

Christian

> ---
>  virt/kvm/eventfd.c |    8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
> index abe4d60..29c2a04 100644
> --- a/virt/kvm/eventfd.c
> +++ b/virt/kvm/eventfd.c
> @@ -391,19 +391,19 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
>  					   lockdep_is_held(&kvm->irqfds.lock));
>  	irqfd_update(kvm, irqfd, irq_rt);
> 
> -	events = f.file->f_op->poll(f.file, &irqfd->pt);
> -
>  	list_add_tail(&irqfd->list, &kvm->irqfds.items);
> 
> +	spin_unlock_irq(&kvm->irqfds.lock);
> +
>  	/*
>  	 * Check if there was an event already pending on the eventfd
>  	 * before we registered, and trigger it as if we didn't miss it.
>  	 */
> +	events = f.file->f_op->poll(f.file, &irqfd->pt);
> +
>  	if (events & POLLIN)
>  		schedule_work(&irqfd->inject);
> 
> -	spin_unlock_irq(&kvm->irqfds.lock);
> -
>  	/*
>  	 * do not drop the file until the irqfd is fully initialized, otherwise
>  	 * we might race against the POLLHUP
> 

Re: [Qemu-devel] [PATCH v2 1/5] KVM: eventfd: Fix lock order inversion.

[Cornelia Huck]

On Mon, 17 Mar 2014 22:55:49 +0100
Christian Borntraeger <borntraeger@de.ibm.com> wrote:

> On 17/03/14 19:11, Cornelia Huck wrote:
> > When registering a new irqfd, we call its ->poll method to collect any
> > event that might have previously been pending so that we can trigger it.
> > This is done under the kvm->irqfds.lock, which means the eventfd's ctx
> > lock is taken under it.
> > 
> > However, if we get a POLLHUP in irqfd_wakeup, we will be called with the
> > ctx lock held before getting the irqfds.lock to deactivate the irqfd,
> > causing lockdep to complain.
> > 
> > Calling the ->poll method does not really need the irqfds.lock, so let's
> > just move it after we've given up the irqfds.lock in kvm_irqfd_assign().
> > 
> > Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> Do you still have the lockdep message somewhere? 

Unfortunately not, and it does not seem to be readily triggerable.

> Looking at the patch and the description this makes sense. Even without
> irqfd for s390:
> Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
> 
> Paolo, maybe this patch can go in independently from s390?
> 
> Christian

Re: [Qemu-devel] [PATCH v2 1/5] KVM: eventfd: Fix lock order inversion.

[Paolo Bonzini]

Il 17/03/2014 22:55, Christian Borntraeger ha scritto:
>> > Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> Do you still have the lockdep message somewhere?
> Looking at the patch and the description this makes sense. Even without
> irqfd for s390:
> Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
>
> Paolo, maybe this patch can go in independently from s390?

Yes, that's fine.  I'm applying it  to kvm/queue.

Paolo

[Qemu-devel] [PATCH v2 2/5] KVM: Add per-vm capability enablement.

[Cornelia Huck]

Allow KVM_ENABLE_CAP to act on a vm as well as on a vcpu. This makes more
sense when the caller wants to enable a vm-related capability.

s390 will be the first user; wire it up.

Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
 Documentation/virtual/kvm/api.txt |    6 ++++--
 arch/s390/kvm/kvm-s390.c          |   24 ++++++++++++++++++++++++
 include/uapi/linux/kvm.h          |    5 +++++
 3 files changed, 33 insertions(+), 2 deletions(-)

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index 4714f28..faf6fe9 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -932,9 +932,9 @@ documentation when it pops into existence).
 
 4.37 KVM_ENABLE_CAP
 
-Capability: KVM_CAP_ENABLE_CAP
+Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
 Architectures: ppc, s390
-Type: vcpu ioctl
+Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
 Parameters: struct kvm_enable_cap (in)
 Returns: 0 on success; -1 on error
 
@@ -965,6 +965,8 @@ function properly, this is the place to put them.
        __u8  pad[64];
 };
 
+The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
+for vm-wide capabilities.
 
 4.38 KVM_GET_MP_STATE
 
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 7337c57..9f1e99f 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -159,6 +159,7 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_S390_CSS_SUPPORT:
 	case KVM_CAP_IOEVENTFD:
 	case KVM_CAP_DEVICE_CTRL:
+	case KVM_CAP_ENABLE_CAP_VM:
 		r = 1;
 		break;
 	case KVM_CAP_NR_VCPUS:
@@ -187,6 +188,21 @@ int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
 	return 0;
 }
 
+static int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
+{
+	int r;
+
+	if (cap->flags)
+		return -EINVAL;
+
+	switch (cap->cap) {
+	default:
+		r = -EINVAL;
+		break;
+	}
+	return r;
+}
+
 long kvm_arch_vm_ioctl(struct file *filp,
 		       unsigned int ioctl, unsigned long arg)
 {
@@ -204,6 +220,14 @@ long kvm_arch_vm_ioctl(struct file *filp,
 		r = kvm_s390_inject_vm(kvm, &s390int);
 		break;
 	}
+	case KVM_ENABLE_CAP: {
+		struct kvm_enable_cap cap;
+		r = -EFAULT;
+		if (copy_from_user(&cap, argp, sizeof(cap)))
+			break;
+		r = kvm_vm_ioctl_enable_cap(kvm, &cap);
+		break;
+	}
 	default:
 		r = -ENOTTY;
 	}
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index a7518be..46ea1b4 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -741,6 +741,7 @@ struct kvm_ppc_smmu_info {
 #define KVM_CAP_EXT_EMUL_CPUID 95
 #define KVM_CAP_HYPERV_TIME 96
 #define KVM_CAP_IOAPIC_POLARITY_IGNORED 97
+#define KVM_CAP_ENABLE_CAP_VM 98
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -1076,6 +1077,10 @@ struct kvm_s390_ucas_mapping {
 /* Available with KVM_CAP_DEBUGREGS */
 #define KVM_GET_DEBUGREGS         _IOR(KVMIO,  0xa1, struct kvm_debugregs)
 #define KVM_SET_DEBUGREGS         _IOW(KVMIO,  0xa2, struct kvm_debugregs)
+/*
+ * vcpu version available with KVM_ENABLE_CAP
+ * vm version available with KVM_CAP_ENABLE_CAP_VM
+ */
 #define KVM_ENABLE_CAP            _IOW(KVMIO,  0xa3, struct kvm_enable_cap)
 /* Available with KVM_CAP_XSAVE */
 #define KVM_GET_XSAVE		  _IOR(KVMIO,  0xa4, struct kvm_xsave)
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v2 2/5] KVM: Add per-vm capability enablement.

[Christian Borntraeger]

On 17/03/14 19:11, Cornelia Huck wrote:
> Allow KVM_ENABLE_CAP to act on a vm as well as on a vcpu. This makes more
> sense when the caller wants to enable a vm-related capability.
> 
> s390 will be the first user; wire it up.
> 
> Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>

> ---
>  Documentation/virtual/kvm/api.txt |    6 ++++--
>  arch/s390/kvm/kvm-s390.c          |   24 ++++++++++++++++++++++++
>  include/uapi/linux/kvm.h          |    5 +++++
>  3 files changed, 33 insertions(+), 2 deletions(-)
> 
> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
> index 4714f28..faf6fe9 100644
> --- a/Documentation/virtual/kvm/api.txt
> +++ b/Documentation/virtual/kvm/api.txt
> @@ -932,9 +932,9 @@ documentation when it pops into existence).
> 
>  4.37 KVM_ENABLE_CAP
> 
> -Capability: KVM_CAP_ENABLE_CAP
> +Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
>  Architectures: ppc, s390
> -Type: vcpu ioctl
> +Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
>  Parameters: struct kvm_enable_cap (in)
>  Returns: 0 on success; -1 on error
> 
> @@ -965,6 +965,8 @@ function properly, this is the place to put them.
>         __u8  pad[64];
>  };
> 
> +The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
> +for vm-wide capabilities.
> 
>  4.38 KVM_GET_MP_STATE
> 
> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
> index 7337c57..9f1e99f 100644
> --- a/arch/s390/kvm/kvm-s390.c
> +++ b/arch/s390/kvm/kvm-s390.c
> @@ -159,6 +159,7 @@ int kvm_dev_ioctl_check_extension(long ext)
>  	case KVM_CAP_S390_CSS_SUPPORT:
>  	case KVM_CAP_IOEVENTFD:
>  	case KVM_CAP_DEVICE_CTRL:
> +	case KVM_CAP_ENABLE_CAP_VM:
>  		r = 1;
>  		break;
>  	case KVM_CAP_NR_VCPUS:
> @@ -187,6 +188,21 @@ int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
>  	return 0;
>  }
> 
> +static int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
> +{
> +	int r;
> +
> +	if (cap->flags)
> +		return -EINVAL;
> +
> +	switch (cap->cap) {
> +	default:
> +		r = -EINVAL;
> +		break;
> +	}
> +	return r;
> +}
> +
>  long kvm_arch_vm_ioctl(struct file *filp,
>  		       unsigned int ioctl, unsigned long arg)
>  {
> @@ -204,6 +220,14 @@ long kvm_arch_vm_ioctl(struct file *filp,
>  		r = kvm_s390_inject_vm(kvm, &s390int);
>  		break;
>  	}
> +	case KVM_ENABLE_CAP: {
> +		struct kvm_enable_cap cap;
> +		r = -EFAULT;
> +		if (copy_from_user(&cap, argp, sizeof(cap)))
> +			break;
> +		r = kvm_vm_ioctl_enable_cap(kvm, &cap);
> +		break;
> +	}
>  	default:
>  		r = -ENOTTY;
>  	}
> diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
> index a7518be..46ea1b4 100644
> --- a/include/uapi/linux/kvm.h
> +++ b/include/uapi/linux/kvm.h
> @@ -741,6 +741,7 @@ struct kvm_ppc_smmu_info {
>  #define KVM_CAP_EXT_EMUL_CPUID 95
>  #define KVM_CAP_HYPERV_TIME 96
>  #define KVM_CAP_IOAPIC_POLARITY_IGNORED 97
> +#define KVM_CAP_ENABLE_CAP_VM 98
> 
>  #ifdef KVM_CAP_IRQ_ROUTING
> 
> @@ -1076,6 +1077,10 @@ struct kvm_s390_ucas_mapping {
>  /* Available with KVM_CAP_DEBUGREGS */
>  #define KVM_GET_DEBUGREGS         _IOR(KVMIO,  0xa1, struct kvm_debugregs)
>  #define KVM_SET_DEBUGREGS         _IOW(KVMIO,  0xa2, struct kvm_debugregs)
> +/*
> + * vcpu version available with KVM_ENABLE_CAP
> + * vm version available with KVM_CAP_ENABLE_CAP_VM
> + */
>  #define KVM_ENABLE_CAP            _IOW(KVMIO,  0xa3, struct kvm_enable_cap)
>  /* Available with KVM_CAP_XSAVE */
>  #define KVM_GET_XSAVE		  _IOR(KVMIO,  0xa4, struct kvm_xsave)
> 

Re: [Qemu-devel] [PATCH v2 2/5] KVM: Add per-vm capability enablement.

[Christian Borntraeger]

On 17/03/14 19:11, Cornelia Huck wrote:
> Allow KVM_ENABLE_CAP to act on a vm as well as on a vcpu. This makes more
> sense when the caller wants to enable a vm-related capability.
> 
> s390 will be the first user; wire it up.
> 
> Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>

> ---
>  Documentation/virtual/kvm/api.txt |    6 ++++--
>  arch/s390/kvm/kvm-s390.c          |   24 ++++++++++++++++++++++++
>  include/uapi/linux/kvm.h          |    5 +++++
>  3 files changed, 33 insertions(+), 2 deletions(-)
> 
> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
> index 4714f28..faf6fe9 100644
> --- a/Documentation/virtual/kvm/api.txt
> +++ b/Documentation/virtual/kvm/api.txt
> @@ -932,9 +932,9 @@ documentation when it pops into existence).
> 
>  4.37 KVM_ENABLE_CAP
> 
> -Capability: KVM_CAP_ENABLE_CAP
> +Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
>  Architectures: ppc, s390
> -Type: vcpu ioctl
> +Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
>  Parameters: struct kvm_enable_cap (in)
>  Returns: 0 on success; -1 on error
> 
> @@ -965,6 +965,8 @@ function properly, this is the place to put them.
>         __u8  pad[64];
>  };
> 
> +The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
> +for vm-wide capabilities.
> 
>  4.38 KVM_GET_MP_STATE
> 
> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
> index 7337c57..9f1e99f 100644
> --- a/arch/s390/kvm/kvm-s390.c
> +++ b/arch/s390/kvm/kvm-s390.c
> @@ -159,6 +159,7 @@ int kvm_dev_ioctl_check_extension(long ext)
>  	case KVM_CAP_S390_CSS_SUPPORT:
>  	case KVM_CAP_IOEVENTFD:
>  	case KVM_CAP_DEVICE_CTRL:
> +	case KVM_CAP_ENABLE_CAP_VM:
>  		r = 1;
>  		break;
>  	case KVM_CAP_NR_VCPUS:
> @@ -187,6 +188,21 @@ int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
>  	return 0;
>  }
> 
> +static int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
> +{
> +	int r;
> +
> +	if (cap->flags)
> +		return -EINVAL;
> +
> +	switch (cap->cap) {
> +	default:
> +		r = -EINVAL;
> +		break;
> +	}
> +	return r;
> +}
> +
>  long kvm_arch_vm_ioctl(struct file *filp,
>  		       unsigned int ioctl, unsigned long arg)
>  {
> @@ -204,6 +220,14 @@ long kvm_arch_vm_ioctl(struct file *filp,
>  		r = kvm_s390_inject_vm(kvm, &s390int);
>  		break;
>  	}
> +	case KVM_ENABLE_CAP: {
> +		struct kvm_enable_cap cap;
> +		r = -EFAULT;
> +		if (copy_from_user(&cap, argp, sizeof(cap)))
> +			break;
> +		r = kvm_vm_ioctl_enable_cap(kvm, &cap);
> +		break;
> +	}
>  	default:
>  		r = -ENOTTY;
>  	}
> diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
> index a7518be..46ea1b4 100644
> --- a/include/uapi/linux/kvm.h
> +++ b/include/uapi/linux/kvm.h
> @@ -741,6 +741,7 @@ struct kvm_ppc_smmu_info {
>  #define KVM_CAP_EXT_EMUL_CPUID 95
>  #define KVM_CAP_HYPERV_TIME 96
>  #define KVM_CAP_IOAPIC_POLARITY_IGNORED 97
> +#define KVM_CAP_ENABLE_CAP_VM 98
> 
>  #ifdef KVM_CAP_IRQ_ROUTING
> 
> @@ -1076,6 +1077,10 @@ struct kvm_s390_ucas_mapping {
>  /* Available with KVM_CAP_DEBUGREGS */
>  #define KVM_GET_DEBUGREGS         _IOR(KVMIO,  0xa1, struct kvm_debugregs)
>  #define KVM_SET_DEBUGREGS         _IOW(KVMIO,  0xa2, struct kvm_debugregs)
> +/*
> + * vcpu version available with KVM_ENABLE_CAP
> + * vm version available with KVM_CAP_ENABLE_CAP_VM
> + */
>  #define KVM_ENABLE_CAP            _IOW(KVMIO,  0xa3, struct kvm_enable_cap)
>  /* Available with KVM_CAP_XSAVE */
>  #define KVM_GET_XSAVE		  _IOR(KVMIO,  0xa4, struct kvm_xsave)
> 

[Qemu-devel] [PATCH v2 3/5] KVM: s390: adapter interrupt sources

[Cornelia Huck]

Add a new interface to register/deregister sources of adapter interrupts
identified by an unique id via the flic. Adapters may also be maskable
and carry a list of pinned pages.

These adapters will be used by irq routing later.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
 Documentation/virtual/kvm/devices/s390_flic.txt |   43 ++++++
 arch/s390/include/asm/kvm_host.h                |   20 +++
 arch/s390/include/uapi/asm/kvm.h                |   22 ++++
 arch/s390/kvm/interrupt.c                       |  161 ++++++++++++++++++++++-
 arch/s390/kvm/kvm-s390.c                        |    1 +
 arch/s390/kvm/kvm-s390.h                        |    2 +
 6 files changed, 248 insertions(+), 1 deletion(-)

diff --git a/Documentation/virtual/kvm/devices/s390_flic.txt b/Documentation/virtual/kvm/devices/s390_flic.txt
index 410fa67..db16111 100644
--- a/Documentation/virtual/kvm/devices/s390_flic.txt
+++ b/Documentation/virtual/kvm/devices/s390_flic.txt
@@ -12,6 +12,7 @@ FLIC provides support to
 - inspect currently pending interrupts (KVM_FLIC_GET_ALL_IRQS)
 - purge all pending floating interrupts (KVM_DEV_FLIC_CLEAR_IRQS)
 - enable/disable for the guest transparent async page faults
+- register and modify adapter interrupt sources (KVM_DEV_FLIC_ADAPTER_*)
 
 Groups:
   KVM_DEV_FLIC_ENQUEUE
@@ -44,3 +45,45 @@ Groups:
     Disables async page faults for the guest and waits until already pending
     async page faults are done. This is necessary to trigger a completion interrupt
     for every init interrupt before migrating the interrupt list.
+
+  KVM_DEV_FLIC_ADAPTER_REGISTER
+    Register an I/O adapter interrupt source. Takes a kvm_s390_io_adapter
+    describing the adapter to register:
+
+struct kvm_s390_io_adapter {
+	__u32 id;
+	__u8 isc;
+	__u8 maskable;
+	__u8 swap;
+	__u8 pad;
+};
+
+   id contains the unique id for the adapter, isc the I/O interruption subclass
+   to use, maskable whether this adapter may be masked (interrupts turned off)
+   and swap whether the indicators need to be byte swapped.
+
+
+  KVM_DEV_FLIC_ADAPTER_MODIFY
+    Modifies attributes of an existing I/O adapter interrupt source. Takes
+    a kvm_s390_io_adapter_req specifiying the adapter and the operation:
+
+struct kvm_s390_io_adapter_req {
+	__u32 id;
+	__u8 type;
+	__u8 mask;
+	__u16 pad0;
+	__u64 addr;
+};
+
+    id specifies the adapter and type the operation. The supported operations
+    are:
+
+    KVM_S390_IO_ADAPTER_MASK
+      mask or unmask the adapter, as specified in mask
+
+    KVM_S390_IO_ADAPTER_MAP
+      pin a userspace page for the address provided in addr and add it to the
+      list of mappings
+
+    KVM_S390_IO_ADAPTER_UNMAP
+      release a userspace page as specified in addr from the list of mappings
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
index 734d302..356f595 100644
--- a/arch/s390/include/asm/kvm_host.h
+++ b/arch/s390/include/asm/kvm_host.h
@@ -19,6 +19,7 @@
 #include <linux/kvm.h>
 #include <asm/debug.h>
 #include <asm/cpu.h>
+#include <asm/isc.h>
 
 #define KVM_MAX_VCPUS 64
 #define KVM_USER_MEM_SLOTS 32
@@ -245,6 +246,24 @@ struct kvm_vm_stat {
 struct kvm_arch_memory_slot {
 };
 
+struct s390_map_info {
+	struct list_head list;
+	__u64 addr;
+	struct page *page;
+};
+
+struct s390_io_adapter {
+	unsigned int id;
+	int isc;
+	bool maskable;
+	bool masked;
+	bool swap;
+	struct rw_semaphore maps_lock;
+	struct list_head maps;
+};
+
+#define MAX_S390_IO_ADAPTERS (MAX_ISC + 1) * 8
+
 struct kvm_arch{
 	struct sca_block *sca;
 	debug_info_t *dbf;
@@ -252,6 +271,7 @@ struct kvm_arch{
 	struct kvm_device *flic;
 	struct gmap *gmap;
 	int css_support;
+	struct s390_io_adapter *adapters[MAX_S390_IO_ADAPTERS];
 };
 
 #define KVM_HVA_ERR_BAD		(-1UL)
diff --git a/arch/s390/include/uapi/asm/kvm.h b/arch/s390/include/uapi/asm/kvm.h
index 2f0ade2..c003c6a 100644
--- a/arch/s390/include/uapi/asm/kvm.h
+++ b/arch/s390/include/uapi/asm/kvm.h
@@ -22,6 +22,8 @@
 #define KVM_DEV_FLIC_CLEAR_IRQS		3
 #define KVM_DEV_FLIC_APF_ENABLE		4
 #define KVM_DEV_FLIC_APF_DISABLE_WAIT	5
+#define KVM_DEV_FLIC_ADAPTER_REGISTER	6
+#define KVM_DEV_FLIC_ADAPTER_MODIFY	7
 /*
  * We can have up to 4*64k pending subchannels + 8 adapter interrupts,
  * as well as up  to ASYNC_PF_PER_VCPU*KVM_MAX_VCPUS pfault done interrupts.
@@ -32,6 +34,26 @@
 #define KVM_S390_MAX_FLOAT_IRQS	266250
 #define KVM_S390_FLIC_MAX_BUFFER	0x2000000
 
+struct kvm_s390_io_adapter {
+	__u32 id;
+	__u8 isc;
+	__u8 maskable;
+	__u8 swap;
+	__u8 pad;
+};
+
+#define KVM_S390_IO_ADAPTER_MASK 1
+#define KVM_S390_IO_ADAPTER_MAP 2
+#define KVM_S390_IO_ADAPTER_UNMAP 3
+
+struct kvm_s390_io_adapter_req {
+	__u32 id;
+	__u8 type;
+	__u8 mask;
+	__u16 pad0;
+	__u64 addr;
+};
+
 /* for KVM_GET_REGS and KVM_SET_REGS */
 struct kvm_regs {
 	/* general purpose regs for s390 */
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index 79d2e4f..94b337e 100644
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -1,7 +1,7 @@
 /*
  * handling kvm guest interrupts
  *
- * Copyright IBM Corp. 2008
+ * Copyright IBM Corp. 2008,2014
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License (version 2 only)
@@ -1054,6 +1054,159 @@ static int enqueue_floating_irq(struct kvm_device *dev,
 	return r;
 }
 
+static struct s390_io_adapter *get_io_adapter(struct kvm *kvm, unsigned int id)
+{
+	if (id >= MAX_S390_IO_ADAPTERS)
+		return NULL;
+	return kvm->arch.adapters[id];
+}
+
+static int register_io_adapter(struct kvm_device *dev,
+			       struct kvm_device_attr *attr)
+{
+	struct s390_io_adapter *adapter;
+	struct kvm_s390_io_adapter adapter_info;
+
+	if (copy_from_user(&adapter_info,
+			   (void __user *)attr->addr, sizeof(adapter_info)))
+		return -EFAULT;
+
+	if ((adapter_info.id >= MAX_S390_IO_ADAPTERS) ||
+	    (dev->kvm->arch.adapters[adapter_info.id] != NULL))
+		return -EINVAL;
+
+	adapter = kzalloc(sizeof(*adapter), GFP_KERNEL);
+	if (!adapter)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&adapter->maps);
+	init_rwsem(&adapter->maps_lock);
+	adapter->id = adapter_info.id;
+	adapter->isc = adapter_info.isc;
+	adapter->maskable = adapter_info.maskable;
+	adapter->masked = false;
+	adapter->swap = adapter_info.swap;
+	dev->kvm->arch.adapters[adapter->id] = adapter;
+
+	return 0;
+}
+
+int kvm_s390_mask_adapter(struct kvm *kvm, unsigned int id, bool masked)
+{
+	int ret;
+	struct s390_io_adapter *adapter = get_io_adapter(kvm, id);
+
+	if (!adapter || !adapter->maskable)
+		return -EINVAL;
+	ret = adapter->masked;
+	adapter->masked = masked;
+	return ret;
+}
+
+static int kvm_s390_adapter_map(struct kvm *kvm, unsigned int id, __u64 addr)
+{
+	struct s390_io_adapter *adapter = get_io_adapter(kvm, id);
+	struct s390_map_info *map;
+	int ret;
+
+	if (!adapter || !addr)
+		return -EINVAL;
+
+	map = kzalloc(sizeof(*map), GFP_KERNEL);
+	if (!map) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	INIT_LIST_HEAD(&map->list);
+	map->addr = addr;
+	ret = get_user_pages_fast(addr, 1, 1, &map->page);
+	if (ret < 0)
+		goto out;
+	BUG_ON(ret != 1);
+	down_write(&adapter->maps_lock);
+	list_add_tail(&map->list, &adapter->maps);
+	up_write(&adapter->maps_lock);
+	ret = 0;
+out:
+	if (ret)
+		kfree(map);
+	return ret;
+}
+
+static int kvm_s390_adapter_unmap(struct kvm *kvm, unsigned int id, __u64 addr)
+{
+	struct s390_io_adapter *adapter = get_io_adapter(kvm, id);
+	struct s390_map_info *map, *tmp;
+	int found = 0;
+
+	if (!adapter || !addr)
+		return -EINVAL;
+
+	down_write(&adapter->maps_lock);
+	list_for_each_entry_safe(map, tmp, &adapter->maps, list) {
+		if (map->addr == addr) {
+			found = 1;
+			list_del(&map->list);
+			put_page(map->page);
+			kfree(map);
+			break;
+		}
+	}
+	up_write(&adapter->maps_lock);
+
+	return found ? 0 : -EINVAL;
+}
+
+void kvm_s390_destroy_adapters(struct kvm *kvm)
+{
+	int i;
+	struct s390_map_info *map, *tmp;
+
+	for (i = 0; i < MAX_S390_IO_ADAPTERS; i++) {
+		if (!kvm->arch.adapters[i])
+			continue;
+		list_for_each_entry_safe(map, tmp,
+					 &kvm->arch.adapters[i]->maps, list) {
+			list_del(&map->list);
+			put_page(map->page);
+			kfree(map);
+		}
+		kfree(kvm->arch.adapters[i]);
+	}
+}
+
+static int modify_io_adapter(struct kvm_device *dev,
+			     struct kvm_device_attr *attr)
+{
+	struct kvm_s390_io_adapter_req req;
+	struct s390_io_adapter *adapter;
+	int ret;
+
+	if (copy_from_user(&req, (void __user *)attr->addr, sizeof(req)))
+		return -EFAULT;
+
+	adapter = get_io_adapter(dev->kvm, req.id);
+	if (!adapter)
+		return -EINVAL;
+	switch (req.type) {
+	case KVM_S390_IO_ADAPTER_MASK:
+		ret = kvm_s390_mask_adapter(dev->kvm, req.id, req.mask);
+		if (ret > 0)
+			ret = 0;
+		break;
+	case KVM_S390_IO_ADAPTER_MAP:
+		ret = kvm_s390_adapter_map(dev->kvm, req.id, req.addr);
+		break;
+	case KVM_S390_IO_ADAPTER_UNMAP:
+		ret = kvm_s390_adapter_unmap(dev->kvm, req.id, req.addr);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
 static int flic_set_attr(struct kvm_device *dev, struct kvm_device_attr *attr)
 {
 	int r = 0;
@@ -1082,6 +1235,12 @@ static int flic_set_attr(struct kvm_device *dev, struct kvm_device_attr *attr)
 		kvm_for_each_vcpu(i, vcpu, dev->kvm)
 			kvm_clear_async_pf_completion_queue(vcpu);
 		break;
+	case KVM_DEV_FLIC_ADAPTER_REGISTER:
+		r = register_io_adapter(dev, attr);
+		break;
+	case KVM_DEV_FLIC_ADAPTER_MODIFY:
+		r = modify_io_adapter(dev, attr);
+		break;
 	default:
 		r = -EINVAL;
 	}
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 9f1e99f..2e6fbb0 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -343,6 +343,7 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
 	debug_unregister(kvm->arch.dbf);
 	if (!kvm_is_ucontrol(kvm))
 		gmap_free(kvm->arch.gmap);
+	kvm_s390_destroy_adapters(kvm);
 }
 
 /* Section: vcpu related */
diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h
index ed4750a..5502cc95 100644
--- a/arch/s390/kvm/kvm-s390.h
+++ b/arch/s390/kvm/kvm-s390.h
@@ -136,6 +136,7 @@ int __must_check kvm_s390_inject_vcpu(struct kvm_vcpu *vcpu,
 int __must_check kvm_s390_inject_program_int(struct kvm_vcpu *vcpu, u16 code);
 struct kvm_s390_interrupt_info *kvm_s390_get_io_int(struct kvm *kvm,
 						    u64 cr6, u64 schid);
+int kvm_s390_mask_adapter(struct kvm *kvm, unsigned int id, bool masked);
 
 /* implemented in priv.c */
 int kvm_s390_handle_b2(struct kvm_vcpu *vcpu);
@@ -162,5 +163,6 @@ int kvm_s390_handle_diag(struct kvm_vcpu *vcpu);
 /* implemented in interrupt.c */
 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
 int psw_extint_disabled(struct kvm_vcpu *vcpu);
+void kvm_s390_destroy_adapters(struct kvm *kvm);
 
 #endif
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v2 3/5] KVM: s390: adapter interrupt sources

[Heiko Carstens]

On Mon, Mar 17, 2014 at 07:11:37PM +0100, Cornelia Huck wrote:
> Add a new interface to register/deregister sources of adapter interrupts
> identified by an unique id via the flic. Adapters may also be maskable
> and carry a list of pinned pages.
> 
> These adapters will be used by irq routing later.
> 
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> ---

[...]

> +#define MAX_S390_IO_ADAPTERS (MAX_ISC + 1) * 8

Subtle possible bug alert ;) Please add braces around (MAX_ISC + 1) * 8.

Re: [Qemu-devel] [PATCH v2 3/5] KVM: s390: adapter interrupt sources

[Cornelia Huck]

On Tue, 18 Mar 2014 09:11:19 +0100
Heiko Carstens <heiko.carstens@de.ibm.com> wrote:

> On Mon, Mar 17, 2014 at 07:11:37PM +0100, Cornelia Huck wrote:
> > Add a new interface to register/deregister sources of adapter interrupts
> > identified by an unique id via the flic. Adapters may also be maskable
> > and carry a list of pinned pages.
> > 
> > These adapters will be used by irq routing later.
> > 
> > Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> > ---
> 
> [...]
> 
> > +#define MAX_S390_IO_ADAPTERS (MAX_ISC + 1) * 8
> 
> Subtle possible bug alert ;) Please add braces around (MAX_ISC + 1) * 8.

Makes sense, will do.

Re: [Qemu-devel] [PATCH v2 3/5] KVM: s390: adapter interrupt sources

[Christian Borntraeger]

On 17/03/14 19:11, Cornelia Huck wrote:
> Add a new interface to register/deregister sources of adapter interrupts
> identified by an unique id via the flic. Adapters may also be maskable
> and carry a list of pinned pages.
> 
> These adapters will be used by irq routing later.
> 
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

[...]

> +static int kvm_s390_adapter_map(struct kvm *kvm, unsigned int id, __u64 addr)
> +{
> +	struct s390_io_adapter *adapter = get_io_adapter(kvm, id);
> +	struct s390_map_info *map;
> +	int ret;
> +
> +	if (!adapter || !addr)
> +		return -EINVAL;
> +
> +	map = kzalloc(sizeof(*map), GFP_KERNEL);
> +	if (!map) {
> +		ret = -ENOMEM;
> +		goto out;
> +	}
> +	INIT_LIST_HEAD(&map->list);
> +	map->addr = addr;
> +	ret = get_user_pages_fast(addr, 1, 1, &map->page);
> +	if (ret < 0)
> +		goto out;
> +	BUG_ON(ret != 1);
> +	down_write(&adapter->maps_lock);
> +	list_add_tail(&map->list, &adapter->maps);
> +	up_write(&adapter->maps_lock);
> +	ret = 0;

Can you limit the amount of pinned pages to something sane, e.g. 1MB?
As far as I can see, QEMU will fall back to non-irqfd if the mapping fails.

Otherwise looks good.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>

Re: [Qemu-devel] [PATCH v2 3/5] KVM: s390: adapter interrupt sources

[Cornelia Huck]

On Fri, 21 Mar 2014 10:26:35 +0100
Christian Borntraeger <borntraeger@de.ibm.com> wrote:

> On 17/03/14 19:11, Cornelia Huck wrote:
> > Add a new interface to register/deregister sources of adapter interrupts
> > identified by an unique id via the flic. Adapters may also be maskable
> > and carry a list of pinned pages.
> > 
> > These adapters will be used by irq routing later.
> > 
> > Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> 
> [...]
> 
> > +static int kvm_s390_adapter_map(struct kvm *kvm, unsigned int id, __u64 addr)
> > +{
> > +	struct s390_io_adapter *adapter = get_io_adapter(kvm, id);
> > +	struct s390_map_info *map;
> > +	int ret;
> > +
> > +	if (!adapter || !addr)
> > +		return -EINVAL;
> > +
> > +	map = kzalloc(sizeof(*map), GFP_KERNEL);
> > +	if (!map) {
> > +		ret = -ENOMEM;
> > +		goto out;
> > +	}
> > +	INIT_LIST_HEAD(&map->list);
> > +	map->addr = addr;
> > +	ret = get_user_pages_fast(addr, 1, 1, &map->page);
> > +	if (ret < 0)
> > +		goto out;
> > +	BUG_ON(ret != 1);
> > +	down_write(&adapter->maps_lock);
> > +	list_add_tail(&map->list, &adapter->maps);
> > +	up_write(&adapter->maps_lock);
> > +	ret = 0;
> 
> Can you limit the amount of pinned pages to something sane, e.g. 1MB?
> As far as I can see, QEMU will fall back to non-irqfd if the mapping fails.

I'll introduce a counter for that.

> 
> Otherwise looks good.
> 
> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
> 

[Qemu-devel] [PATCH v2 4/5] KVM: s390: irq routing for adapter interrupts.

[Cornelia Huck]

Introduce a new interrupt class for s390 adapter interrupts and enable
irqfds for s390.

This is depending on a new s390 specific vm capability, KVM_CAP_S390_IRQCHIP,
that needs to be enabled by userspace.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
 Documentation/virtual/kvm/api.txt               |   21 +++-
 Documentation/virtual/kvm/devices/s390_flic.txt |    6 +-
 arch/s390/include/asm/kvm_host.h                |   10 ++
 arch/s390/kvm/Kconfig                           |    2 +
 arch/s390/kvm/Makefile                          |    2 +-
 arch/s390/kvm/interrupt.c                       |  132 ++++++++++++++++++++++-
 arch/s390/kvm/irq.h                             |   22 ++++
 arch/s390/kvm/kvm-s390.c                        |   17 +++
 include/linux/kvm_host.h                        |    9 ++
 include/uapi/linux/kvm.h                        |   11 ++
 10 files changed, 222 insertions(+), 10 deletions(-)
 create mode 100644 arch/s390/kvm/irq.h

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index faf6fe9..2cb1640 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -586,8 +586,8 @@ struct kvm_fpu {
 
 4.24 KVM_CREATE_IRQCHIP
 
-Capability: KVM_CAP_IRQCHIP
-Architectures: x86, ia64, ARM, arm64
+Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
+Architectures: x86, ia64, ARM, arm64, s390
 Type: vm ioctl
 Parameters: none
 Returns: 0 on success, -1 on error
@@ -596,7 +596,10 @@ Creates an interrupt controller model in the kernel.  On x86, creates a virtual
 ioapic, a virtual PIC (two PICs, nested), and sets up future vcpus to have a
 local APIC.  IRQ routing for GSIs 0-15 is set to both PIC and IOAPIC; GSI 16-23
 only go to the IOAPIC.  On ia64, a IOSAPIC is created. On ARM/arm64, a GIC is
-created.
+created. On s390, a dummy irq routing table is created.
+
+Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
+before KVM_CREATE_IRQCHIP can be used.
 
 
 4.25 KVM_IRQ_LINE
@@ -1336,7 +1339,7 @@ KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
 4.52 KVM_SET_GSI_ROUTING
 
 Capability: KVM_CAP_IRQ_ROUTING
-Architectures: x86 ia64
+Architectures: x86 ia64 s390
 Type: vm ioctl
 Parameters: struct kvm_irq_routing (in)
 Returns: 0 on success, -1 on error
@@ -1359,6 +1362,7 @@ struct kvm_irq_routing_entry {
 	union {
 		struct kvm_irq_routing_irqchip irqchip;
 		struct kvm_irq_routing_msi msi;
+		struct kvm_irq_routing_s390_adapter adapter;
 		__u32 pad[8];
 	} u;
 };
@@ -1366,6 +1370,7 @@ struct kvm_irq_routing_entry {
 /* gsi routing entry types */
 #define KVM_IRQ_ROUTING_IRQCHIP 1
 #define KVM_IRQ_ROUTING_MSI 2
+#define KVM_IRQ_ROUTING_S390_ADAPTER 3
 
 No flags are specified so far, the corresponding field must be set to zero.
 
@@ -1381,6 +1386,14 @@ struct kvm_irq_routing_msi {
 	__u32 pad;
 };
 
+struct kvm_irq_routing_s390_adapter {
+	__u64 ind_addr;
+	__u64 summary_addr;
+	__u64 ind_offset;
+	__u32 summary_offset;
+	__u32 adapter_id;
+};
+
 
 4.53 KVM_ASSIGN_SET_MSIX_NR
 
diff --git a/Documentation/virtual/kvm/devices/s390_flic.txt b/Documentation/virtual/kvm/devices/s390_flic.txt
index db16111..4ceef53 100644
--- a/Documentation/virtual/kvm/devices/s390_flic.txt
+++ b/Documentation/virtual/kvm/devices/s390_flic.txt
@@ -82,8 +82,10 @@ struct kvm_s390_io_adapter_req {
       mask or unmask the adapter, as specified in mask
 
     KVM_S390_IO_ADAPTER_MAP
-      pin a userspace page for the address provided in addr and add it to the
+      perform a gmap translation for the guest address provided in addr,
+      pin a userspace page for the translated address and add it to the
       list of mappings
 
     KVM_S390_IO_ADAPTER_UNMAP
-      release a userspace page as specified in addr from the list of mappings
+      release a userspace page for the translated address specified in addr
+      from the list of mappings
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
index 356f595..82bf697 100644
--- a/arch/s390/include/asm/kvm_host.h
+++ b/arch/s390/include/asm/kvm_host.h
@@ -24,6 +24,14 @@
 #define KVM_MAX_VCPUS 64
 #define KVM_USER_MEM_SLOTS 32
 
+/*
+ * These seem to be used for allocating ->chip in the routing table,
+ * which we don't use. 4096 is an out-of-thin-air value. If we need
+ * to look at ->chip later on, we'll need to revisit this.
+ */
+#define KVM_NR_IRQCHIPS 1
+#define KVM_IRQCHIP_NUM_PINS 4096
+
 struct sca_entry {
 	atomic_t scn;
 	__u32	reserved;
@@ -248,6 +256,7 @@ struct kvm_arch_memory_slot {
 
 struct s390_map_info {
 	struct list_head list;
+	__u64 guest_addr;
 	__u64 addr;
 	struct page *page;
 };
@@ -271,6 +280,7 @@ struct kvm_arch{
 	struct kvm_device *flic;
 	struct gmap *gmap;
 	int css_support;
+	int use_irqchip;
 	struct s390_io_adapter *adapters[MAX_S390_IO_ADAPTERS];
 };
 
diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
index c8bacbc..10d529a 100644
--- a/arch/s390/kvm/Kconfig
+++ b/arch/s390/kvm/Kconfig
@@ -25,6 +25,8 @@ config KVM
 	select HAVE_KVM_EVENTFD
 	select KVM_ASYNC_PF
 	select KVM_ASYNC_PF_SYNC
+	select HAVE_KVM_IRQCHIP
+	select HAVE_KVM_IRQ_ROUTING
 	---help---
 	  Support hosting paravirtualized guest machines using the SIE
 	  virtualization capability on the mainframe. This should work
diff --git a/arch/s390/kvm/Makefile b/arch/s390/kvm/Makefile
index a47d2c3..d3adb37 100644
--- a/arch/s390/kvm/Makefile
+++ b/arch/s390/kvm/Makefile
@@ -7,7 +7,7 @@
 # as published by the Free Software Foundation.
 
 KVM := ../../../virt/kvm
-common-objs = $(KVM)/kvm_main.o $(KVM)/eventfd.o  $(KVM)/async_pf.o
+common-objs = $(KVM)/kvm_main.o $(KVM)/eventfd.o  $(KVM)/async_pf.o $(KVM)/irqchip.o
 
 ccflags-y := -Ivirt/kvm -Iarch/s390/kvm
 
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index 94b337e..8155bb4 100644
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -13,6 +13,7 @@
 #include <linux/interrupt.h>
 #include <linux/kvm_host.h>
 #include <linux/hrtimer.h>
+#include <linux/mmu_context.h>
 #include <linux/signal.h>
 #include <linux/slab.h>
 #include <asm/asm-offsets.h>
@@ -1118,8 +1119,13 @@ static int kvm_s390_adapter_map(struct kvm *kvm, unsigned int id, __u64 addr)
 		goto out;
 	}
 	INIT_LIST_HEAD(&map->list);
-	map->addr = addr;
-	ret = get_user_pages_fast(addr, 1, 1, &map->page);
+	map->guest_addr = addr;
+	map->addr = gmap_translate(addr, kvm->arch.gmap);
+	if (map->addr == -EFAULT) {
+		ret = -EFAULT;
+		goto out;
+	}
+	ret = get_user_pages_fast(map->addr, 1, 1, &map->page);
 	if (ret < 0)
 		goto out;
 	BUG_ON(ret != 1);
@@ -1144,7 +1150,7 @@ static int kvm_s390_adapter_unmap(struct kvm *kvm, unsigned int id, __u64 addr)
 
 	down_write(&adapter->maps_lock);
 	list_for_each_entry_safe(map, tmp, &adapter->maps, list) {
-		if (map->addr == addr) {
+		if (map->guest_addr == addr) {
 			found = 1;
 			list_del(&map->list);
 			put_page(map->page);
@@ -1272,3 +1278,123 @@ struct kvm_device_ops kvm_flic_ops = {
 	.create = flic_create,
 	.destroy = flic_destroy,
 };
+
+static unsigned long get_ind_bit(__u64 addr, unsigned long bit_nr, bool swap)
+{
+	unsigned long bit;
+
+	bit = bit_nr + (addr % PAGE_SIZE) * 8;
+
+	return swap ? (bit ^ (BITS_PER_LONG - 1)) : bit;
+}
+
+static struct s390_map_info *get_map_info(struct s390_io_adapter *adapter,
+					  u64 addr)
+{
+	struct s390_map_info *map;
+
+	if (!adapter)
+		return NULL;
+
+	list_for_each_entry(map, &adapter->maps, list) {
+		if (map->guest_addr == addr)
+			return map;
+	}
+	return NULL;
+}
+
+static int adapter_indicators_set(struct kvm *kvm,
+				  struct s390_io_adapter *adapter,
+				  struct kvm_s390_adapter_int *adapter_int)
+{
+	unsigned long bit;
+	int summary_set, idx;
+	struct s390_map_info *info;
+	void *map;
+
+	info = get_map_info(adapter, adapter_int->ind_addr);
+	if (!info)
+		return -1;
+	map = page_address(info->page);
+	bit = get_ind_bit(info->addr, adapter_int->ind_offset, adapter->swap);
+	set_bit(bit, map);
+	idx = srcu_read_lock(&kvm->srcu);
+	mark_page_dirty(kvm, info->guest_addr >> PAGE_SHIFT);
+	set_page_dirty_lock(info->page);
+	info = get_map_info(adapter, adapter_int->summary_addr);
+	if (!info) {
+		srcu_read_unlock(&kvm->srcu, idx);
+		return -1;
+	}
+	map = page_address(info->page);
+	bit = get_ind_bit(info->addr, adapter_int->summary_offset,
+			  adapter->swap);
+	summary_set = test_and_set_bit(bit, map);
+	mark_page_dirty(kvm, info->guest_addr >> PAGE_SHIFT);
+	set_page_dirty_lock(info->page);
+	srcu_read_unlock(&kvm->srcu, idx);
+	return summary_set ? 0 : 1;
+}
+
+/*
+ * < 0 - not injected due to error
+ * = 0 - coalesced, summary indicator already active
+ * > 0 - injected interrupt
+ */
+static int set_adapter_int(struct kvm_kernel_irq_routing_entry *e,
+			   struct kvm *kvm, int irq_source_id, int level,
+			   bool line_status)
+{
+	int ret;
+	struct s390_io_adapter *adapter;
+
+	/* We're only interested in the 0->1 transition. */
+	if (!level)
+		return 0;
+	adapter = get_io_adapter(kvm, e->adapter.adapter_id);
+	if (!adapter)
+		return -1;
+	down_read(&adapter->maps_lock);
+	ret = adapter_indicators_set(kvm, adapter, &e->adapter);
+	up_read(&adapter->maps_lock);
+	if ((ret > 0) && !adapter->masked) {
+		struct kvm_s390_interrupt s390int = {
+			.type = KVM_S390_INT_IO(1, 0, 0, 0),
+			.parm = 0,
+			.parm64 = (adapter->isc << 27) | 0x80000000,
+		};
+		ret = kvm_s390_inject_vm(kvm, &s390int);
+		if (ret == 0)
+			ret = 1;
+	}
+	return ret;
+}
+
+int kvm_set_routing_entry(struct kvm_irq_routing_table *rt,
+			  struct kvm_kernel_irq_routing_entry *e,
+			  const struct kvm_irq_routing_entry *ue)
+{
+	int ret;
+
+	switch (ue->type) {
+	case KVM_IRQ_ROUTING_S390_ADAPTER:
+		e->set = set_adapter_int;
+		e->adapter.summary_addr = ue->u.adapter.summary_addr;
+		e->adapter.ind_addr = ue->u.adapter.ind_addr;
+		e->adapter.summary_offset = ue->u.adapter.summary_offset;
+		e->adapter.ind_offset = ue->u.adapter.ind_offset;
+		e->adapter.adapter_id = ue->u.adapter.adapter_id;
+		ret = 0;
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
+int kvm_set_msi(struct kvm_kernel_irq_routing_entry *e, struct kvm *kvm,
+		int irq_source_id, int level, bool line_status)
+{
+	return -EINVAL;
+}
diff --git a/arch/s390/kvm/irq.h b/arch/s390/kvm/irq.h
new file mode 100644
index 0000000..d98e415
--- /dev/null
+++ b/arch/s390/kvm/irq.h
@@ -0,0 +1,22 @@
+/*
+ * s390 irqchip routines
+ *
+ * Copyright IBM Corp. 2014
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License (version 2 only)
+ * as published by the Free Software Foundation.
+ *
+ *    Author(s): Cornelia Huck <cornelia.huck@de.ibm.com>
+ */
+#ifndef __KVM_IRQ_H
+#define __KVM_IRQ_H
+
+#include <linux/kvm_host.h>
+
+static inline int irqchip_in_kernel(struct kvm *kvm)
+{
+	return 1;
+}
+
+#endif
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 2e6fbb0..ce5b659 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -196,6 +196,10 @@ static int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
 		return -EINVAL;
 
 	switch (cap->cap) {
+	case KVM_CAP_S390_IRQCHIP:
+		kvm->arch.use_irqchip = 1;
+		r = 0;
+		break;
 	default:
 		r = -EINVAL;
 		break;
@@ -228,6 +232,18 @@ long kvm_arch_vm_ioctl(struct file *filp,
 		r = kvm_vm_ioctl_enable_cap(kvm, &cap);
 		break;
 	}
+	case KVM_CREATE_IRQCHIP: {
+		struct kvm_irq_routing_entry routing;
+
+		r = -EINVAL;
+		if (kvm->arch.use_irqchip) {
+			/* Set up dummy routing. */
+			memset(&routing, 0, sizeof(routing));
+			kvm_set_irq_routing(kvm, &routing, 0, 0);
+			r = 0;
+		}
+		break;
+	}
 	default:
 		r = -ENOTTY;
 	}
@@ -284,6 +300,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 	}
 
 	kvm->arch.css_support = 0;
+	kvm->arch.use_irqchip = 0;
 
 	return 0;
 out_nogmap:
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 9816b68..da7510b 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -297,6 +297,14 @@ static inline unsigned long kvm_dirty_bitmap_bytes(struct kvm_memory_slot *memsl
 	return ALIGN(memslot->npages, BITS_PER_LONG) / 8;
 }
 
+struct kvm_s390_adapter_int {
+	u64 ind_addr;
+	u64 summary_addr;
+	u64 ind_offset;
+	u32 summary_offset;
+	u32 adapter_id;
+};
+
 struct kvm_kernel_irq_routing_entry {
 	u32 gsi;
 	u32 type;
@@ -309,6 +317,7 @@ struct kvm_kernel_irq_routing_entry {
 			unsigned pin;
 		} irqchip;
 		struct msi_msg msi;
+		struct kvm_s390_adapter_int adapter;
 	};
 	struct hlist_node link;
 };
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index 46ea1b4..a8f4ee5 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -742,6 +742,7 @@ struct kvm_ppc_smmu_info {
 #define KVM_CAP_HYPERV_TIME 96
 #define KVM_CAP_IOAPIC_POLARITY_IGNORED 97
 #define KVM_CAP_ENABLE_CAP_VM 98
+#define KVM_CAP_S390_IRQCHIP 99
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -757,9 +758,18 @@ struct kvm_irq_routing_msi {
 	__u32 pad;
 };
 
+struct kvm_irq_routing_s390_adapter {
+	__u64 ind_addr;
+	__u64 summary_addr;
+	__u64 ind_offset;
+	__u32 summary_offset;
+	__u32 adapter_id;
+};
+
 /* gsi routing entry types */
 #define KVM_IRQ_ROUTING_IRQCHIP 1
 #define KVM_IRQ_ROUTING_MSI 2
+#define KVM_IRQ_ROUTING_S390_ADAPTER 3
 
 struct kvm_irq_routing_entry {
 	__u32 gsi;
@@ -769,6 +779,7 @@ struct kvm_irq_routing_entry {
 	union {
 		struct kvm_irq_routing_irqchip irqchip;
 		struct kvm_irq_routing_msi msi;
+		struct kvm_irq_routing_s390_adapter adapter;
 		__u32 pad[8];
 	} u;
 };
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v2 4/5] KVM: s390: irq routing for adapter interrupts.

[Christian Borntraeger]

On 17/03/14 19:11, Cornelia Huck wrote:
> Introduce a new interrupt class for s390 adapter interrupts and enable
> irqfds for s390.
> 
> This is depending on a new s390 specific vm capability, KVM_CAP_S390_IRQCHIP,
> that needs to be enabled by userspace.
> 
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> ---
>  Documentation/virtual/kvm/api.txt               |   21 +++-
>  Documentation/virtual/kvm/devices/s390_flic.txt |    6 +-
>  arch/s390/include/asm/kvm_host.h                |   10 ++
>  arch/s390/kvm/Kconfig                           |    2 +
>  arch/s390/kvm/Makefile                          |    2 +-
>  arch/s390/kvm/interrupt.c                       |  132 ++++++++++++++++++++++-
>  arch/s390/kvm/irq.h                             |   22 ++++
>  arch/s390/kvm/kvm-s390.c                        |   17 +++
>  include/linux/kvm_host.h                        |    9 ++
>  include/uapi/linux/kvm.h                        |   11 ++
>  10 files changed, 222 insertions(+), 10 deletions(-)
>  create mode 100644 arch/s390/kvm/irq.h
> 
> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
> index faf6fe9..2cb1640 100644
> --- a/Documentation/virtual/kvm/api.txt
> +++ b/Documentation/virtual/kvm/api.txt
> @@ -586,8 +586,8 @@ struct kvm_fpu {
> 
>  4.24 KVM_CREATE_IRQCHIP
> 
> -Capability: KVM_CAP_IRQCHIP
> -Architectures: x86, ia64, ARM, arm64
> +Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
> +Architectures: x86, ia64, ARM, arm64, s390
>  Type: vm ioctl
>  Parameters: none
>  Returns: 0 on success, -1 on error
> @@ -596,7 +596,10 @@ Creates an interrupt controller model in the kernel.  On x86, creates a virtual
>  ioapic, a virtual PIC (two PICs, nested), and sets up future vcpus to have a
>  local APIC.  IRQ routing for GSIs 0-15 is set to both PIC and IOAPIC; GSI 16-23
>  only go to the IOAPIC.  On ia64, a IOSAPIC is created. On ARM/arm64, a GIC is
> -created.
> +created. On s390, a dummy irq routing table is created.
> +
> +Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
> +before KVM_CREATE_IRQCHIP can be used.
> 
> 
>  4.25 KVM_IRQ_LINE
> @@ -1336,7 +1339,7 @@ KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
>  4.52 KVM_SET_GSI_ROUTING
> 
>  Capability: KVM_CAP_IRQ_ROUTING
> -Architectures: x86 ia64
> +Architectures: x86 ia64 s390
>  Type: vm ioctl
>  Parameters: struct kvm_irq_routing (in)
>  Returns: 0 on success, -1 on error
> @@ -1359,6 +1362,7 @@ struct kvm_irq_routing_entry {
>  	union {
>  		struct kvm_irq_routing_irqchip irqchip;
>  		struct kvm_irq_routing_msi msi;
> +		struct kvm_irq_routing_s390_adapter adapter;
>  		__u32 pad[8];
>  	} u;
>  };
> @@ -1366,6 +1370,7 @@ struct kvm_irq_routing_entry {
>  /* gsi routing entry types */
>  #define KVM_IRQ_ROUTING_IRQCHIP 1
>  #define KVM_IRQ_ROUTING_MSI 2
> +#define KVM_IRQ_ROUTING_S390_ADAPTER 3
> 
>  No flags are specified so far, the corresponding field must be set to zero.
> 
> @@ -1381,6 +1386,14 @@ struct kvm_irq_routing_msi {
>  	__u32 pad;
>  };
> 
> +struct kvm_irq_routing_s390_adapter {
> +	__u64 ind_addr;
> +	__u64 summary_addr;
> +	__u64 ind_offset;
> +	__u32 summary_offset;
> +	__u32 adapter_id;
> +};
> +
> 
>  4.53 KVM_ASSIGN_SET_MSIX_NR
> 
> diff --git a/Documentation/virtual/kvm/devices/s390_flic.txt b/Documentation/virtual/kvm/devices/s390_flic.txt
> index db16111..4ceef53 100644
> --- a/Documentation/virtual/kvm/devices/s390_flic.txt
> +++ b/Documentation/virtual/kvm/devices/s390_flic.txt
> @@ -82,8 +82,10 @@ struct kvm_s390_io_adapter_req {
>        mask or unmask the adapter, as specified in mask
> 
>      KVM_S390_IO_ADAPTER_MAP
> -      pin a userspace page for the address provided in addr and add it to the
> +      perform a gmap translation for the guest address provided in addr,
> +      pin a userspace page for the translated address and add it to the
>        list of mappings
> 
>      KVM_S390_IO_ADAPTER_UNMAP
> -      release a userspace page as specified in addr from the list of mappings
> +      release a userspace page for the translated address specified in addr
> +      from the list of mappings
> diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
> index 356f595..82bf697 100644
> --- a/arch/s390/include/asm/kvm_host.h
> +++ b/arch/s390/include/asm/kvm_host.h
> @@ -24,6 +24,14 @@
>  #define KVM_MAX_VCPUS 64
>  #define KVM_USER_MEM_SLOTS 32
> 
> +/*
> + * These seem to be used for allocating ->chip in the routing table,
> + * which we don't use. 4096 is an out-of-thin-air value. If we need
> + * to look at ->chip later on, we'll need to revisit this.
> + */
> +#define KVM_NR_IRQCHIPS 1
> +#define KVM_IRQCHIP_NUM_PINS 4096
> +
>  struct sca_entry {
>  	atomic_t scn;
>  	__u32	reserved;
> @@ -248,6 +256,7 @@ struct kvm_arch_memory_slot {
> 
>  struct s390_map_info {
>  	struct list_head list;
> +	__u64 guest_addr;
>  	__u64 addr;
>  	struct page *page;
>  };
> @@ -271,6 +280,7 @@ struct kvm_arch{
>  	struct kvm_device *flic;
>  	struct gmap *gmap;
>  	int css_support;
> +	int use_irqchip;
>  	struct s390_io_adapter *adapters[MAX_S390_IO_ADAPTERS];
>  };
> 
> diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
> index c8bacbc..10d529a 100644
> --- a/arch/s390/kvm/Kconfig
> +++ b/arch/s390/kvm/Kconfig
> @@ -25,6 +25,8 @@ config KVM
>  	select HAVE_KVM_EVENTFD
>  	select KVM_ASYNC_PF
>  	select KVM_ASYNC_PF_SYNC
> +	select HAVE_KVM_IRQCHIP
> +	select HAVE_KVM_IRQ_ROUTING
>  	---help---
>  	  Support hosting paravirtualized guest machines using the SIE
>  	  virtualization capability on the mainframe. This should work
> diff --git a/arch/s390/kvm/Makefile b/arch/s390/kvm/Makefile
> index a47d2c3..d3adb37 100644
> --- a/arch/s390/kvm/Makefile
> +++ b/arch/s390/kvm/Makefile
> @@ -7,7 +7,7 @@
>  # as published by the Free Software Foundation.
> 
>  KVM := ../../../virt/kvm
> -common-objs = $(KVM)/kvm_main.o $(KVM)/eventfd.o  $(KVM)/async_pf.o
> +common-objs = $(KVM)/kvm_main.o $(KVM)/eventfd.o  $(KVM)/async_pf.o $(KVM)/irqchip.o
> 
>  ccflags-y := -Ivirt/kvm -Iarch/s390/kvm
> 
> diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
> index 94b337e..8155bb4 100644
> --- a/arch/s390/kvm/interrupt.c
> +++ b/arch/s390/kvm/interrupt.c
> @@ -13,6 +13,7 @@
>  #include <linux/interrupt.h>
>  #include <linux/kvm_host.h>
>  #include <linux/hrtimer.h>
> +#include <linux/mmu_context.h>
>  #include <linux/signal.h>
>  #include <linux/slab.h>
>  #include <asm/asm-offsets.h>
> @@ -1118,8 +1119,13 @@ static int kvm_s390_adapter_map(struct kvm *kvm, unsigned int id, __u64 addr)
>  		goto out;
>  	}
>  	INIT_LIST_HEAD(&map->list);
> -	map->addr = addr;
> -	ret = get_user_pages_fast(addr, 1, 1, &map->page);
> +	map->guest_addr = addr;
> +	map->addr = gmap_translate(addr, kvm->arch.gmap);
> +	if (map->addr == -EFAULT) {
> +		ret = -EFAULT;
> +		goto out;
> +	}
> +	ret = get_user_pages_fast(map->addr, 1, 1, &map->page);
>  	if (ret < 0)
>  		goto out;
>  	BUG_ON(ret != 1);
> @@ -1144,7 +1150,7 @@ static int kvm_s390_adapter_unmap(struct kvm *kvm, unsigned int id, __u64 addr)
> 
>  	down_write(&adapter->maps_lock);
>  	list_for_each_entry_safe(map, tmp, &adapter->maps, list) {
> -		if (map->addr == addr) {
> +		if (map->guest_addr == addr) {
>  			found = 1;
>  			list_del(&map->list);
>  			put_page(map->page);


Can't these two hunks be merged into the previous patch? 

Otherwise:
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>



> @@ -1272,3 +1278,123 @@ struct kvm_device_ops kvm_flic_ops = {
>  	.create = flic_create,
>  	.destroy = flic_destroy,
>  };
> +
> +static unsigned long get_ind_bit(__u64 addr, unsigned long bit_nr, bool swap)
> +{
> +	unsigned long bit;
> +
> +	bit = bit_nr + (addr % PAGE_SIZE) * 8;
> +
> +	return swap ? (bit ^ (BITS_PER_LONG - 1)) : bit;
> +}
> +
> +static struct s390_map_info *get_map_info(struct s390_io_adapter *adapter,
> +					  u64 addr)
> +{
> +	struct s390_map_info *map;
> +
> +	if (!adapter)
> +		return NULL;
> +
> +	list_for_each_entry(map, &adapter->maps, list) {
> +		if (map->guest_addr == addr)
> +			return map;
> +	}
> +	return NULL;
> +}
> +
> +static int adapter_indicators_set(struct kvm *kvm,
> +				  struct s390_io_adapter *adapter,
> +				  struct kvm_s390_adapter_int *adapter_int)
> +{
> +	unsigned long bit;
> +	int summary_set, idx;
> +	struct s390_map_info *info;
> +	void *map;
> +
> +	info = get_map_info(adapter, adapter_int->ind_addr);
> +	if (!info)
> +		return -1;
> +	map = page_address(info->page);
> +	bit = get_ind_bit(info->addr, adapter_int->ind_offset, adapter->swap);
> +	set_bit(bit, map);
> +	idx = srcu_read_lock(&kvm->srcu);
> +	mark_page_dirty(kvm, info->guest_addr >> PAGE_SHIFT);
> +	set_page_dirty_lock(info->page);
> +	info = get_map_info(adapter, adapter_int->summary_addr);
> +	if (!info) {
> +		srcu_read_unlock(&kvm->srcu, idx);
> +		return -1;
> +	}
> +	map = page_address(info->page);
> +	bit = get_ind_bit(info->addr, adapter_int->summary_offset,
> +			  adapter->swap);
> +	summary_set = test_and_set_bit(bit, map);
> +	mark_page_dirty(kvm, info->guest_addr >> PAGE_SHIFT);
> +	set_page_dirty_lock(info->page);
> +	srcu_read_unlock(&kvm->srcu, idx);
> +	return summary_set ? 0 : 1;
> +}
> +
> +/*
> + * < 0 - not injected due to error
> + * = 0 - coalesced, summary indicator already active
> + * > 0 - injected interrupt
> + */
> +static int set_adapter_int(struct kvm_kernel_irq_routing_entry *e,
> +			   struct kvm *kvm, int irq_source_id, int level,
> +			   bool line_status)
> +{
> +	int ret;
> +	struct s390_io_adapter *adapter;
> +
> +	/* We're only interested in the 0->1 transition. */
> +	if (!level)
> +		return 0;
> +	adapter = get_io_adapter(kvm, e->adapter.adapter_id);
> +	if (!adapter)
> +		return -1;
> +	down_read(&adapter->maps_lock);
> +	ret = adapter_indicators_set(kvm, adapter, &e->adapter);
> +	up_read(&adapter->maps_lock);
> +	if ((ret > 0) && !adapter->masked) {
> +		struct kvm_s390_interrupt s390int = {
> +			.type = KVM_S390_INT_IO(1, 0, 0, 0),
> +			.parm = 0,
> +			.parm64 = (adapter->isc << 27) | 0x80000000,
> +		};
> +		ret = kvm_s390_inject_vm(kvm, &s390int);
> +		if (ret == 0)
> +			ret = 1;
> +	}
> +	return ret;
> +}
> +
> +int kvm_set_routing_entry(struct kvm_irq_routing_table *rt,
> +			  struct kvm_kernel_irq_routing_entry *e,
> +			  const struct kvm_irq_routing_entry *ue)
> +{
> +	int ret;
> +
> +	switch (ue->type) {
> +	case KVM_IRQ_ROUTING_S390_ADAPTER:
> +		e->set = set_adapter_int;
> +		e->adapter.summary_addr = ue->u.adapter.summary_addr;
> +		e->adapter.ind_addr = ue->u.adapter.ind_addr;
> +		e->adapter.summary_offset = ue->u.adapter.summary_offset;
> +		e->adapter.ind_offset = ue->u.adapter.ind_offset;
> +		e->adapter.adapter_id = ue->u.adapter.adapter_id;
> +		ret = 0;
> +		break;
> +	default:
> +		ret = -EINVAL;
> +	}
> +
> +	return ret;
> +}
> +
> +int kvm_set_msi(struct kvm_kernel_irq_routing_entry *e, struct kvm *kvm,
> +		int irq_source_id, int level, bool line_status)
> +{
> +	return -EINVAL;
> +}
> diff --git a/arch/s390/kvm/irq.h b/arch/s390/kvm/irq.h
> new file mode 100644
> index 0000000..d98e415
> --- /dev/null
> +++ b/arch/s390/kvm/irq.h
> @@ -0,0 +1,22 @@
> +/*
> + * s390 irqchip routines
> + *
> + * Copyright IBM Corp. 2014
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License (version 2 only)
> + * as published by the Free Software Foundation.
> + *
> + *    Author(s): Cornelia Huck <cornelia.huck@de.ibm.com>
> + */
> +#ifndef __KVM_IRQ_H
> +#define __KVM_IRQ_H
> +
> +#include <linux/kvm_host.h>
> +
> +static inline int irqchip_in_kernel(struct kvm *kvm)
> +{
> +	return 1;
> +}
> +
> +#endif
> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
> index 2e6fbb0..ce5b659 100644
> --- a/arch/s390/kvm/kvm-s390.c
> +++ b/arch/s390/kvm/kvm-s390.c
> @@ -196,6 +196,10 @@ static int kvm_vm_ioctl_enable_cap(struct kvm *kvm, struct kvm_enable_cap *cap)
>  		return -EINVAL;
> 
>  	switch (cap->cap) {
> +	case KVM_CAP_S390_IRQCHIP:
> +		kvm->arch.use_irqchip = 1;
> +		r = 0;
> +		break;
>  	default:
>  		r = -EINVAL;
>  		break;
> @@ -228,6 +232,18 @@ long kvm_arch_vm_ioctl(struct file *filp,
>  		r = kvm_vm_ioctl_enable_cap(kvm, &cap);
>  		break;
>  	}
> +	case KVM_CREATE_IRQCHIP: {
> +		struct kvm_irq_routing_entry routing;
> +
> +		r = -EINVAL;
> +		if (kvm->arch.use_irqchip) {
> +			/* Set up dummy routing. */
> +			memset(&routing, 0, sizeof(routing));
> +			kvm_set_irq_routing(kvm, &routing, 0, 0);
> +			r = 0;
> +		}
> +		break;
> +	}
>  	default:
>  		r = -ENOTTY;
>  	}
> @@ -284,6 +300,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
>  	}
> 
>  	kvm->arch.css_support = 0;
> +	kvm->arch.use_irqchip = 0;
> 
>  	return 0;
>  out_nogmap:
> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index 9816b68..da7510b 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -297,6 +297,14 @@ static inline unsigned long kvm_dirty_bitmap_bytes(struct kvm_memory_slot *memsl
>  	return ALIGN(memslot->npages, BITS_PER_LONG) / 8;
>  }
> 
> +struct kvm_s390_adapter_int {
> +	u64 ind_addr;
> +	u64 summary_addr;
> +	u64 ind_offset;
> +	u32 summary_offset;
> +	u32 adapter_id;
> +};
> +
>  struct kvm_kernel_irq_routing_entry {
>  	u32 gsi;
>  	u32 type;
> @@ -309,6 +317,7 @@ struct kvm_kernel_irq_routing_entry {
>  			unsigned pin;
>  		} irqchip;
>  		struct msi_msg msi;
> +		struct kvm_s390_adapter_int adapter;
>  	};
>  	struct hlist_node link;
>  };
> diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
> index 46ea1b4..a8f4ee5 100644
> --- a/include/uapi/linux/kvm.h
> +++ b/include/uapi/linux/kvm.h
> @@ -742,6 +742,7 @@ struct kvm_ppc_smmu_info {
>  #define KVM_CAP_HYPERV_TIME 96
>  #define KVM_CAP_IOAPIC_POLARITY_IGNORED 97
>  #define KVM_CAP_ENABLE_CAP_VM 98
> +#define KVM_CAP_S390_IRQCHIP 99
> 
>  #ifdef KVM_CAP_IRQ_ROUTING
> 
> @@ -757,9 +758,18 @@ struct kvm_irq_routing_msi {
>  	__u32 pad;
>  };
> 
> +struct kvm_irq_routing_s390_adapter {
> +	__u64 ind_addr;
> +	__u64 summary_addr;
> +	__u64 ind_offset;
> +	__u32 summary_offset;
> +	__u32 adapter_id;
> +};
> +
>  /* gsi routing entry types */
>  #define KVM_IRQ_ROUTING_IRQCHIP 1
>  #define KVM_IRQ_ROUTING_MSI 2
> +#define KVM_IRQ_ROUTING_S390_ADAPTER 3
> 
>  struct kvm_irq_routing_entry {
>  	__u32 gsi;
> @@ -769,6 +779,7 @@ struct kvm_irq_routing_entry {
>  	union {
>  		struct kvm_irq_routing_irqchip irqchip;
>  		struct kvm_irq_routing_msi msi;
> +		struct kvm_irq_routing_s390_adapter adapter;
>  		__u32 pad[8];
>  	} u;
>  };
> 

Re: [Qemu-devel] [PATCH v2 4/5] KVM: s390: irq routing for adapter interrupts.

[Cornelia Huck]

On Fri, 21 Mar 2014 10:32:03 +0100
Christian Borntraeger <borntraeger@de.ibm.com> wrote:

> On 17/03/14 19:11, Cornelia Huck wrote:
> > Introduce a new interrupt class for s390 adapter interrupts and enable
> > irqfds for s390.
> > 
> > This is depending on a new s390 specific vm capability, KVM_CAP_S390_IRQCHIP,
> > that needs to be enabled by userspace.
> > 
> > Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> > ---
> >  Documentation/virtual/kvm/api.txt               |   21 +++-
> >  Documentation/virtual/kvm/devices/s390_flic.txt |    6 +-
> >  arch/s390/include/asm/kvm_host.h                |   10 ++
> >  arch/s390/kvm/Kconfig                           |    2 +
> >  arch/s390/kvm/Makefile                          |    2 +-
> >  arch/s390/kvm/interrupt.c                       |  132 ++++++++++++++++++++++-
> >  arch/s390/kvm/irq.h                             |   22 ++++
> >  arch/s390/kvm/kvm-s390.c                        |   17 +++
> >  include/linux/kvm_host.h                        |    9 ++
> >  include/uapi/linux/kvm.h                        |   11 ++
> >  10 files changed, 222 insertions(+), 10 deletions(-)
> >  create mode 100644 arch/s390/kvm/irq.h

> > diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
> > index 94b337e..8155bb4 100644
> > --- a/arch/s390/kvm/interrupt.c
> > +++ b/arch/s390/kvm/interrupt.c
> > @@ -13,6 +13,7 @@
> >  #include <linux/interrupt.h>
> >  #include <linux/kvm_host.h>
> >  #include <linux/hrtimer.h>
> > +#include <linux/mmu_context.h>
> >  #include <linux/signal.h>
> >  #include <linux/slab.h>
> >  #include <asm/asm-offsets.h>
> > @@ -1118,8 +1119,13 @@ static int kvm_s390_adapter_map(struct kvm *kvm, unsigned int id, __u64 addr)
> >  		goto out;
> >  	}
> >  	INIT_LIST_HEAD(&map->list);
> > -	map->addr = addr;
> > -	ret = get_user_pages_fast(addr, 1, 1, &map->page);
> > +	map->guest_addr = addr;
> > +	map->addr = gmap_translate(addr, kvm->arch.gmap);
> > +	if (map->addr == -EFAULT) {
> > +		ret = -EFAULT;
> > +		goto out;
> > +	}
> > +	ret = get_user_pages_fast(map->addr, 1, 1, &map->page);
> >  	if (ret < 0)
> >  		goto out;
> >  	BUG_ON(ret != 1);
> > @@ -1144,7 +1150,7 @@ static int kvm_s390_adapter_unmap(struct kvm *kvm, unsigned int id, __u64 addr)
> > 
> >  	down_write(&adapter->maps_lock);
> >  	list_for_each_entry_safe(map, tmp, &adapter->maps, list) {
> > -		if (map->addr == addr) {
> > +		if (map->guest_addr == addr) {
> >  			found = 1;
> >  			list_del(&map->list);
> >  			put_page(map->page);
> 
> 
> Can't these two hunks be merged into the previous patch? 

The guest_addr stuff should be in the previous patch, proably some
rebasing fallout. Will fix.

> 
> Otherwise:
> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
> 

[Qemu-devel] [PATCH v2 5/5] KVM: Bump KVM_MAX_IRQ_ROUTES for s390

[Cornelia Huck]

The maximum number for irq routes is currently 1024, which is a bit on
the small size for s390: We support up to 4 x 64k virtual devices with
up to 64 queues, and we need one route for each of the queues if we want
to operate it via irqfd.

Let's bump this to 4k on s390 for now, as this at least covers the saner
setups.

We need to find a more general solution, though, as we can't just grow
the routing table indefinitly.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
 include/linux/kvm_host.h |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index da7510b..7d21cf9 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -922,7 +922,11 @@ static inline int mmu_notifier_retry(struct kvm *kvm, unsigned long mmu_seq)
 
 #ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
 
+#ifdef CONFIG_S390
+#define KVM_MAX_IRQ_ROUTES 4096 //FIXME: we can have more than that...
+#else
 #define KVM_MAX_IRQ_ROUTES 1024
+#endif
 
 int kvm_setup_default_irq_routing(struct kvm *kvm);
 int kvm_set_irq_routing(struct kvm *kvm,
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v2 5/5] KVM: Bump KVM_MAX_IRQ_ROUTES for s390

[Paolo Bonzini]

Il 17/03/2014 19:11, Cornelia Huck ha scritto:
> The maximum number for irq routes is currently 1024, which is a bit on
> the small size for s390: We support up to 4 x 64k virtual devices with
> up to 64 queues, and we need one route for each of the queues if we want
> to operate it via irqfd.
>
> Let's bump this to 4k on s390 for now, as this at least covers the saner
> setups.
>
> We need to find a more general solution, though, as we can't just grow
> the routing table indefinitly.
>
> Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> ---
>  include/linux/kvm_host.h |    4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index da7510b..7d21cf9 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -922,7 +922,11 @@ static inline int mmu_notifier_retry(struct kvm *kvm, unsigned long mmu_seq)
>
>  #ifdef CONFIG_HAVE_KVM_IRQ_ROUTING
>
> +#ifdef CONFIG_S390
> +#define KVM_MAX_IRQ_ROUTES 4096 //FIXME: we can have more than that...
> +#else
>  #define KVM_MAX_IRQ_ROUTES 1024
> +#endif
>
>  int kvm_setup_default_irq_routing(struct kvm *kvm);
>  int kvm_set_irq_routing(struct kvm *kvm,
>

Acked-by: Paolo Bonzini <pbonzini@redhat.com>