[RFC PATCH 24/45] KVM: arm/arm64: vgic-new: Add GICv3 CTLR, IIDR, TYPER handlers

Christoffer Dall christoffer.dall at linaro.org
Tue Apr 12 06:20:07 PDT 2016


On Mon, Apr 11, 2016 at 02:00:02PM +0100, Andre Przywara wrote:
> Hi,
> 
> On 31/03/16 12:53, Christoffer Dall wrote:
> > On Fri, Mar 25, 2016 at 02:04:47AM +0000, Andre Przywara wrote:
> >> As in the GICv2 emulation we handle those three registers in one
> >> function.
> >>
> >> Signed-off-by: Andre Przywara <andre.przywara at arm.com>
> >> ---
> >>  virt/kvm/arm/vgic/vgic.h      |  2 ++
> >>  virt/kvm/arm/vgic/vgic_mmio.c | 38 ++++++++++++++++++++++++++++++++++++--
> >>  2 files changed, 38 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/virt/kvm/arm/vgic/vgic.h b/virt/kvm/arm/vgic/vgic.h
> >> index 4b8952a..0db1abe 100644
> >> --- a/virt/kvm/arm/vgic/vgic.h
> >> +++ b/virt/kvm/arm/vgic/vgic.h
> >> @@ -19,6 +19,8 @@
> >>  #define PRODUCT_ID_KVM		0x4b	/* ASCII code K */
> >>  #define IMPLEMENTER_ARM		0x43b
> >>  
> >> +#define INTERRUPT_ID_BITS_SPIS	10
> >> +
> >>  struct vgic_irq *vgic_get_irq(struct kvm *kvm, struct kvm_vcpu *vcpu,
> >>  			      u32 intid);
> >>  bool vgic_queue_irq(struct kvm *kvm, struct vgic_irq *irq);
> >> diff --git a/virt/kvm/arm/vgic/vgic_mmio.c b/virt/kvm/arm/vgic/vgic_mmio.c
> >> index 2d10c06..13e101f 100644
> >> --- a/virt/kvm/arm/vgic/vgic_mmio.c
> >> +++ b/virt/kvm/arm/vgic/vgic_mmio.c
> >> @@ -606,7 +606,29 @@ static int vgic_mmio_read_v3_misc(struct kvm_vcpu *vcpu,
> >>  				  struct kvm_io_device *this,
> >>  				  gpa_t addr, int len, void *val)
> >>  {
> >> -	/* TODO: implement */
> >> +	struct vgic_io_device *iodev = container_of(this,
> >> +						    struct vgic_io_device, dev);
> >> +	u32 value = 0;
> >> +
> >> +	switch ((addr - iodev->base_addr) & ~3) {
> >> +	case GICD_CTLR:
> >> +		if (vcpu->kvm->arch.vgic.enabled)
> >> +		       value |=	GICD_CTLR_ENABLE_SS_G1;
> >> +		value |= GICD_CTLR_ARE_NS | GICD_CTLR_DS;
> >> +		break;
> >> +	case GICD_TYPER:
> >> +		value = vcpu->kvm->arch.vgic.nr_spis + VGIC_NR_PRIVATE_IRQS;
> >> +		value = (value >> 5) - 1;
> >> +		value |= (INTERRUPT_ID_BITS_SPIS - 1) << 19;
> >> +		break;
> >> +	case GICD_IIDR:
> >> +		value = (PRODUCT_ID_KVM << 24) | (IMPLEMENTER_ARM << 0);
> >> +		break;
> >> +	default:
> >> +		return 0;
> >> +	}
> >> +
> >> +	write_mask32(value, addr & 3, len, val);
> >>  	return 0;
> >>  }
> >>  
> >> @@ -614,7 +636,19 @@ static int vgic_mmio_write_v3_misc(struct kvm_vcpu *vcpu,
> >>  				   struct kvm_io_device *this,
> >>  				   gpa_t addr, int len, const void *val)
> >>  {
> >> -	/* TODO: implement */
> >> +	struct vgic_io_device *iodev = container_of(this,
> >> +						    struct vgic_io_device, dev);
> >> +	bool enabled;
> >> +
> >> +	/* These are not the bits you are looking for ... */
> > 
> > haha, nice.  But I don't understand what this case is?
> 
> Of the 16 bytes this region covers (CTLR, TYPER, IIDR) only the first
> byte is writeable, the rest is WI.
> I am afraid I have to replace the cool quote with something more
> meaningful ...
> 

You can maintain the fun quote and supplement it with a boring
explanation.

> >> +	if (addr - iodev->base_addr > 0)
> >> +		return 0;
> >> +
> >> +	/* We only care about the enable bit, all other bits are WI. */
> >> +	enabled = *(u8*)val & GICD_CTLR_ENABLE_SS_G1;
> >> +
> >> +	vcpu->kvm->arch.vgic.enabled = enabled;
> > 
> > I think some of the comments from the v2 side apply here as well.
> 
> Agreed.
> 

Thanks,
-Christoffer



More information about the linux-arm-kernel mailing list