diff mbox

[v4,01/12] KVM: extend struct kvm_msi to hold a 32-bit device ID

Message ID 1458958450-19662-2-git-send-email-andre.przywara@arm.com (mailing list archive)
State New, archived
Headers show

Commit Message

Andre Przywara March 26, 2016, 2:13 a.m. UTC 
The ARM GICv3 ITS MSI controller requires a device ID to be able to
assign the proper interrupt vector. On real hardware, this ID is
sampled from the bus. To be able to emulate an ITS controller, extend
the KVM MSI interface to let userspace provide such a device ID. For
PCI devices, the device ID is simply the 16-bit bus-device-function
triplet, which should be easily available to the userland tool.

Also there is a new KVM capability which advertises whether the
current VM requires a device ID to be set along with the MSI data.
This flag is still reported as not available everywhere, later we will
enable it when ITS emulation is used.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Eric Auger <eric.auger@linaro.org>
---
 Documentation/virtual/kvm/api.txt | 12 ++++++++++--
 include/uapi/linux/kvm.h          |  5 ++++-
 2 files changed, 14 insertions(+), 3 deletions(-)

Comments

Christoffer Dall April 3, 2016, 9:15 a.m. UTC | #1 
On Sat, Mar 26, 2016 at 02:13:59AM +0000, Andre Przywara wrote:
> The ARM GICv3 ITS MSI controller requires a device ID to be able to
> assign the proper interrupt vector. On real hardware, this ID is
> sampled from the bus. To be able to emulate an ITS controller, extend
> the KVM MSI interface to let userspace provide such a device ID. For
> PCI devices, the device ID is simply the 16-bit bus-device-function
> triplet, which should be easily available to the userland tool.
> 
> Also there is a new KVM capability which advertises whether the
> current VM requires a device ID to be set along with the MSI data.
> This flag is still reported as not available everywhere, later we will
> enable it when ITS emulation is used.
> 
> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
> Reviewed-by: Eric Auger <eric.auger@linaro.org>
> ---
>  Documentation/virtual/kvm/api.txt | 12 ++++++++++--
>  include/uapi/linux/kvm.h          |  5 ++++-
>  2 files changed, 14 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
> index cb2ef0b..8f7351d 100644
> --- a/Documentation/virtual/kvm/api.txt
> +++ b/Documentation/virtual/kvm/api.txt
> @@ -2163,10 +2163,18 @@ struct kvm_msi {
>  	__u32 address_hi;
>  	__u32 data;
>  	__u32 flags;
> -	__u8  pad[16];
> +	__u32 devid;

Are we imposing any unfortunate restrictions for other architectures by
using a u32 over a u64 for the device ID?

-Christoffer

> +	__u8  pad[12];
>  };
>  
> -No flags are defined so far. The corresponding field must be 0.
> +flags: KVM_MSI_VALID_DEVID: devid contains a valid value
> +devid: If KVM_MSI_VALID_DEVID is set, contains a unique device identifier
> +       for the device that wrote the MSI message.
> +       For PCI, this is usually a BFD identifier in the lower 16 bits.
> +
> +The per-VM KVM_CAP_MSI_DEVID capability advertises the need to provide
> +the device ID. If this capability is not set, userland cannot rely on
> +the kernel to allow the KVM_MSI_VALID_DEVID flag being set.
>  
>  
>  4.71 KVM_CREATE_PIT2
> diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
> index 50f44a2..6a02871 100644
> --- a/include/uapi/linux/kvm.h
> +++ b/include/uapi/linux/kvm.h
> @@ -852,6 +852,7 @@ struct kvm_ppc_smmu_info {
>  #define KVM_CAP_S390_RI 124
>  #define KVM_CAP_ARM_PMU_V3 125
>  #define KVM_CAP_VCPU_ATTRIBUTES 126
> +#define KVM_CAP_MSI_DEVID 127
>  
>  #ifdef KVM_CAP_IRQ_ROUTING
>  
> @@ -1010,12 +1011,14 @@ struct kvm_one_reg {
>  	__u64 addr;
>  };
>  
> +#define KVM_MSI_VALID_DEVID	(1U << 0)
>  struct kvm_msi {
>  	__u32 address_lo;
>  	__u32 address_hi;
>  	__u32 data;
>  	__u32 flags;
> -	__u8  pad[16];
> +	__u32 devid;
> +	__u8  pad[12];
>  };
>  
>  struct kvm_arm_device_addr {
> -- 
> 2.7.3
>
Chalamarla, Tirumalesh May 5, 2016, 5:55 p.m. UTC | #2 
On 3/25/16, 7:13 PM, "kvmarm-bounces@lists.cs.columbia.edu on behalf of Andre Przywara" <kvmarm-bounces@lists.cs.columbia.edu on behalf of andre.przywara@arm.com> wrote:

>The ARM GICv3 ITS MSI controller requires a device ID to be able to
>assign the proper interrupt vector. On real hardware, this ID is
>sampled from the bus. To be able to emulate an ITS controller, extend
>the KVM MSI interface to let userspace provide such a device ID. For
>PCI devices, the device ID is simply the 16-bit bus-device-function
>triplet, which should be easily available to the userland tool.

Once VFIO is used it becomes impossible to guess the device ID.
Does the VFIO series take care of this? 
>
>Also there is a new KVM capability which advertises whether the
>current VM requires a device ID to be set along with the MSI data.
>This flag is still reported as not available everywhere, later we will
>enable it when ITS emulation is used.
>
>Signed-off-by: Andre Przywara <andre.przywara@arm.com>
>Reviewed-by: Eric Auger <eric.auger@linaro.org>
>---
> Documentation/virtual/kvm/api.txt | 12 ++++++++++--
> include/uapi/linux/kvm.h          |  5 ++++-
> 2 files changed, 14 insertions(+), 3 deletions(-)
>
>diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
>index cb2ef0b..8f7351d 100644
>--- a/Documentation/virtual/kvm/api.txt
>+++ b/Documentation/virtual/kvm/api.txt
>@@ -2163,10 +2163,18 @@ struct kvm_msi {
> 	__u32 address_hi;
> 	__u32 data;
> 	__u32 flags;
>-	__u8  pad[16];
>+	__u32 devid;
>+	__u8  pad[12];
> };
> 
>-No flags are defined so far. The corresponding field must be 0.
>+flags: KVM_MSI_VALID_DEVID: devid contains a valid value
>+devid: If KVM_MSI_VALID_DEVID is set, contains a unique device identifier
>+       for the device that wrote the MSI message.
>+       For PCI, this is usually a BFD identifier in the lower 16 bits.
>+
>+The per-VM KVM_CAP_MSI_DEVID capability advertises the need to provide
>+the device ID. If this capability is not set, userland cannot rely on
>+the kernel to allow the KVM_MSI_VALID_DEVID flag being set.
> 
> 
> 4.71 KVM_CREATE_PIT2
>diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
>index 50f44a2..6a02871 100644
>--- a/include/uapi/linux/kvm.h
>+++ b/include/uapi/linux/kvm.h
>@@ -852,6 +852,7 @@ struct kvm_ppc_smmu_info {
> #define KVM_CAP_S390_RI 124
> #define KVM_CAP_ARM_PMU_V3 125
> #define KVM_CAP_VCPU_ATTRIBUTES 126
>+#define KVM_CAP_MSI_DEVID 127
> 
> #ifdef KVM_CAP_IRQ_ROUTING
> 
>@@ -1010,12 +1011,14 @@ struct kvm_one_reg {
> 	__u64 addr;
> };
> 
>+#define KVM_MSI_VALID_DEVID	(1U << 0)
> struct kvm_msi {
> 	__u32 address_lo;
> 	__u32 address_hi;
> 	__u32 data;
> 	__u32 flags;
>-	__u8  pad[16];
>+	__u32 devid;
>+	__u8  pad[12];
> };
> 
> struct kvm_arm_device_addr {
>-- 
>2.7.3
>
>_______________________________________________
>kvmarm mailing list
>kvmarm@lists.cs.columbia.edu
>https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Andre Przywara May 25, 2016, 3:55 p.m. UTC | #3 
Hi Christoffer,

On 03/04/16 10:15, Christoffer Dall wrote:
> On Sat, Mar 26, 2016 at 02:13:59AM +0000, Andre Przywara wrote:
>> The ARM GICv3 ITS MSI controller requires a device ID to be able to
>> assign the proper interrupt vector. On real hardware, this ID is
>> sampled from the bus. To be able to emulate an ITS controller, extend
>> the KVM MSI interface to let userspace provide such a device ID. For
>> PCI devices, the device ID is simply the 16-bit bus-device-function
>> triplet, which should be easily available to the userland tool.
>>
>> Also there is a new KVM capability which advertises whether the
>> current VM requires a device ID to be set along with the MSI data.
>> This flag is still reported as not available everywhere, later we will
>> enable it when ITS emulation is used.
>>
>> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
>> Reviewed-by: Eric Auger <eric.auger@linaro.org>
>> ---
>>  Documentation/virtual/kvm/api.txt | 12 ++++++++++--
>>  include/uapi/linux/kvm.h          |  5 ++++-
>>  2 files changed, 14 insertions(+), 3 deletions(-)
>>
>> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
>> index cb2ef0b..8f7351d 100644
>> --- a/Documentation/virtual/kvm/api.txt
>> +++ b/Documentation/virtual/kvm/api.txt
>> @@ -2163,10 +2163,18 @@ struct kvm_msi {
>>  	__u32 address_hi;
>>  	__u32 data;
>>  	__u32 flags;
>> -	__u8  pad[16];
>> +	__u32 devid;
> 
> Are we imposing any unfortunate restrictions for other architectures by
> using a u32 over a u64 for the device ID?

Mmmh, good point. I guess not only for other architectures, but also for
the future in general.

Are there any objections against increasing this to a u64?

Cheers,
Andre.

>> +	__u8  pad[12];
>>  };
>>  
>> -No flags are defined so far. The corresponding field must be 0.
>> +flags: KVM_MSI_VALID_DEVID: devid contains a valid value
>> +devid: If KVM_MSI_VALID_DEVID is set, contains a unique device identifier
>> +       for the device that wrote the MSI message.
>> +       For PCI, this is usually a BFD identifier in the lower 16 bits.
>> +
>> +The per-VM KVM_CAP_MSI_DEVID capability advertises the need to provide
>> +the device ID. If this capability is not set, userland cannot rely on
>> +the kernel to allow the KVM_MSI_VALID_DEVID flag being set.
>>  
>>  
>>  4.71 KVM_CREATE_PIT2
>> diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
>> index 50f44a2..6a02871 100644
>> --- a/include/uapi/linux/kvm.h
>> +++ b/include/uapi/linux/kvm.h
>> @@ -852,6 +852,7 @@ struct kvm_ppc_smmu_info {
>>  #define KVM_CAP_S390_RI 124
>>  #define KVM_CAP_ARM_PMU_V3 125
>>  #define KVM_CAP_VCPU_ATTRIBUTES 126
>> +#define KVM_CAP_MSI_DEVID 127
>>  
>>  #ifdef KVM_CAP_IRQ_ROUTING
>>  
>> @@ -1010,12 +1011,14 @@ struct kvm_one_reg {
>>  	__u64 addr;
>>  };
>>  
>> +#define KVM_MSI_VALID_DEVID	(1U << 0)
>>  struct kvm_msi {
>>  	__u32 address_lo;
>>  	__u32 address_hi;
>>  	__u32 data;
>>  	__u32 flags;
>> -	__u8  pad[16];
>> +	__u32 devid;
>> +	__u8  pad[12];
>>  };
>>  
>>  struct kvm_arm_device_addr {
>> -- 
>> 2.7.3
>>
>
Marc Zyngier May 25, 2016, 4:16 p.m. UTC | #4 
On 25/05/16 16:55, Andre Przywara wrote:
> Hi Christoffer,
> 
> On 03/04/16 10:15, Christoffer Dall wrote:
>> On Sat, Mar 26, 2016 at 02:13:59AM +0000, Andre Przywara wrote:
>>> The ARM GICv3 ITS MSI controller requires a device ID to be able to
>>> assign the proper interrupt vector. On real hardware, this ID is
>>> sampled from the bus. To be able to emulate an ITS controller, extend
>>> the KVM MSI interface to let userspace provide such a device ID. For
>>> PCI devices, the device ID is simply the 16-bit bus-device-function
>>> triplet, which should be easily available to the userland tool.
>>>
>>> Also there is a new KVM capability which advertises whether the
>>> current VM requires a device ID to be set along with the MSI data.
>>> This flag is still reported as not available everywhere, later we will
>>> enable it when ITS emulation is used.
>>>
>>> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
>>> Reviewed-by: Eric Auger <eric.auger@linaro.org>
>>> ---
>>>  Documentation/virtual/kvm/api.txt | 12 ++++++++++--
>>>  include/uapi/linux/kvm.h          |  5 ++++-
>>>  2 files changed, 14 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
>>> index cb2ef0b..8f7351d 100644
>>> --- a/Documentation/virtual/kvm/api.txt
>>> +++ b/Documentation/virtual/kvm/api.txt
>>> @@ -2163,10 +2163,18 @@ struct kvm_msi {
>>>  	__u32 address_hi;
>>>  	__u32 data;
>>>  	__u32 flags;
>>> -	__u8  pad[16];
>>> +	__u32 devid;
>>
>> Are we imposing any unfortunate restrictions for other architectures by
>> using a u32 over a u64 for the device ID?
> 
> Mmmh, good point. I guess not only for other architectures, but also for
> the future in general.
> 
> Are there any objections against increasing this to a u64?

I'm not sure this is really necessary. A PCI RID is 16bit, and expanding
it to 32bit is already making quite a bit of space for further extension.

Also, some architecture might be unhappy of having a 64bit quantity that
is not 64bit aligned...

Thanks,

	M.
Christoffer Dall May 31, 2016, 1:05 p.m. UTC | #5 
On Wed, May 25, 2016 at 05:16:54PM +0100, Marc Zyngier wrote:
> On 25/05/16 16:55, Andre Przywara wrote:
> > Hi Christoffer,
> > 
> > On 03/04/16 10:15, Christoffer Dall wrote:
> >> On Sat, Mar 26, 2016 at 02:13:59AM +0000, Andre Przywara wrote:
> >>> The ARM GICv3 ITS MSI controller requires a device ID to be able to
> >>> assign the proper interrupt vector. On real hardware, this ID is
> >>> sampled from the bus. To be able to emulate an ITS controller, extend
> >>> the KVM MSI interface to let userspace provide such a device ID. For
> >>> PCI devices, the device ID is simply the 16-bit bus-device-function
> >>> triplet, which should be easily available to the userland tool.
> >>>
> >>> Also there is a new KVM capability which advertises whether the
> >>> current VM requires a device ID to be set along with the MSI data.
> >>> This flag is still reported as not available everywhere, later we will
> >>> enable it when ITS emulation is used.
> >>>
> >>> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
> >>> Reviewed-by: Eric Auger <eric.auger@linaro.org>
> >>> ---
> >>>  Documentation/virtual/kvm/api.txt | 12 ++++++++++--
> >>>  include/uapi/linux/kvm.h          |  5 ++++-
> >>>  2 files changed, 14 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
> >>> index cb2ef0b..8f7351d 100644
> >>> --- a/Documentation/virtual/kvm/api.txt
> >>> +++ b/Documentation/virtual/kvm/api.txt
> >>> @@ -2163,10 +2163,18 @@ struct kvm_msi {
> >>>  	__u32 address_hi;
> >>>  	__u32 data;
> >>>  	__u32 flags;
> >>> -	__u8  pad[16];
> >>> +	__u32 devid;
> >>
> >> Are we imposing any unfortunate restrictions for other architectures by
> >> using a u32 over a u64 for the device ID?
> > 
> > Mmmh, good point. I guess not only for other architectures, but also for
> > the future in general.
> > 
> > Are there any objections against increasing this to a u64?
> 
> I'm not sure this is really necessary. A PCI RID is 16bit, and expanding
> it to 32bit is already making quite a bit of space for further extension.
> 
> Also, some architecture might be unhappy of having a 64bit quantity that
> is not 64bit aligned...
> 
ok, thanks for the explanation.

-Christoffer
diff mbox

Patch

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index cb2ef0b..8f7351d 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -2163,10 +2163,18 @@  struct kvm_msi {
 	__u32 address_hi;
 	__u32 data;
 	__u32 flags;
-	__u8  pad[16];
+	__u32 devid;
+	__u8  pad[12];
 };
 
-No flags are defined so far. The corresponding field must be 0.
+flags: KVM_MSI_VALID_DEVID: devid contains a valid value
+devid: If KVM_MSI_VALID_DEVID is set, contains a unique device identifier
+       for the device that wrote the MSI message.
+       For PCI, this is usually a BFD identifier in the lower 16 bits.
+
+The per-VM KVM_CAP_MSI_DEVID capability advertises the need to provide
+the device ID. If this capability is not set, userland cannot rely on
+the kernel to allow the KVM_MSI_VALID_DEVID flag being set.
 
 
 4.71 KVM_CREATE_PIT2
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index 50f44a2..6a02871 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -852,6 +852,7 @@  struct kvm_ppc_smmu_info {
 #define KVM_CAP_S390_RI 124
 #define KVM_CAP_ARM_PMU_V3 125
 #define KVM_CAP_VCPU_ATTRIBUTES 126
+#define KVM_CAP_MSI_DEVID 127
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -1010,12 +1011,14 @@  struct kvm_one_reg {
 	__u64 addr;
 };
 
+#define KVM_MSI_VALID_DEVID	(1U << 0)
 struct kvm_msi {
 	__u32 address_lo;
 	__u32 address_hi;
 	__u32 data;
 	__u32 flags;
-	__u8  pad[16];
+	__u32 devid;
+	__u8  pad[12];
 };
 
 struct kvm_arm_device_addr {