[PATCH v3 24/55] KVM: arm/arm64: vgic-new: Add ENABLE registers handlers

[Andre Przywara]

On 11/05/16 14:14, Christoffer Dall wrote:
> On Wed, May 11, 2016 at 02:04:13PM +0100, Andre Przywara wrote:
>> Hi,
>>
>> On 11/05/16 13:34, Christoffer Dall wrote:
>>> On Fri, May 06, 2016 at 11:45:37AM +0100, Andre Przywara wrote:
>>>> As the enable register handlers are shared between the v2 and v3
>>>> emulation, their implementation goes into vgic-mmio.c, to be easily
>>>> referenced from the v3 emulation as well later.
>>>>
>>>> Signed-off-by: Andre Przywara <andre.przywara at arm.com>
>>>> ---
>>>> Changelog RFC..v1:
>>>> - use lower bits of address to determine IRQ number
>>>> - remove TODO, confirmed to be fine
>>>>
>>>> Changelog v1 .. v2:
>>>> - adapt to new MMIO framework
>>>>
>>>>  virt/kvm/arm/vgic/vgic-mmio-v2.c |  4 +--
>>>>  virt/kvm/arm/vgic/vgic-mmio.c    | 56 ++++++++++++++++++++++++++++++++++++++++
>>>>  virt/kvm/arm/vgic/vgic-mmio.h    | 11 ++++++++
>>>>  3 files changed, 69 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/virt/kvm/arm/vgic/vgic-mmio-v2.c b/virt/kvm/arm/vgic/vgic-mmio-v2.c
>>>> index 69e96f7..448d1da 100644
>>>> --- a/virt/kvm/arm/vgic/vgic-mmio-v2.c
>>>> +++ b/virt/kvm/arm/vgic/vgic-mmio-v2.c
>>>> @@ -72,9 +72,9 @@ static const struct vgic_register_region vgic_v2_dist_registers[] = {
>>>>  	REGISTER_DESC_WITH_BITS_PER_IRQ(GIC_DIST_IGROUP,
>>>>  		vgic_mmio_read_rao, vgic_mmio_write_wi, 1),
>>>>  	REGISTER_DESC_WITH_BITS_PER_IRQ(GIC_DIST_ENABLE_SET,
>>>> -		vgic_mmio_read_raz, vgic_mmio_write_wi, 1),
>>>> +		vgic_mmio_read_enable, vgic_mmio_write_senable, 1),
>>>>  	REGISTER_DESC_WITH_BITS_PER_IRQ(GIC_DIST_ENABLE_CLEAR,
>>>> -		vgic_mmio_read_raz, vgic_mmio_write_wi, 1),
>>>> +		vgic_mmio_read_enable, vgic_mmio_write_cenable, 1),
>>>>  	REGISTER_DESC_WITH_BITS_PER_IRQ(GIC_DIST_PENDING_SET,
>>>>  		vgic_mmio_read_raz, vgic_mmio_write_wi, 1),
>>>>  	REGISTER_DESC_WITH_BITS_PER_IRQ(GIC_DIST_PENDING_CLEAR,
>>>> diff --git a/virt/kvm/arm/vgic/vgic-mmio.c b/virt/kvm/arm/vgic/vgic-mmio.c
>>>> index 41cf4f4..077ae86 100644
>>>> --- a/virt/kvm/arm/vgic/vgic-mmio.c
>>>> +++ b/virt/kvm/arm/vgic/vgic-mmio.c
>>>> @@ -46,6 +46,62 @@ void vgic_mmio_write_wi(struct kvm_vcpu *vcpu, gpa_t addr,
>>>>  	/* Ignore */
>>>>  }
>>>>  
>>>> +/*
>>>> + * Read accesses to both GICD_ICENABLER and GICD_ISENABLER return the value
>>>> + * of the enabled bit, so there is only one function for both here.
>>>> + */
>>>> +unsigned long vgic_mmio_read_enable(struct kvm_vcpu *vcpu,
>>>> +				    gpa_t addr, unsigned int len)
>>>> +{
>>>> +	u32 intid = (addr & 0x7f) * 8;
>>>
>>> is there anything we can do about this to make it more intuitive?  A
>>> macro to generate the mask/offset based on bits per interrupt or
>>> something?
>>
>> Yes, something where you give it the address and the bits-per-IRQ and it
>> tells you the IRQ number.
>> Not sure it is advisable to squash this into v4 still?
>>
>>>
>>>> +	u32 value = 0;
>>>> +	int i;
>>>> +
>>>> +	/* Loop over all IRQs affected by this read */
>>>> +	for (i = 0; i < len * 8; i++) {
>>>> +		struct vgic_irq *irq = vgic_get_irq(vcpu->kvm, vcpu, intid + i);
>>>> +
>>>> +		if (irq->enabled)
>>>> +			value |= (1U << i);
>>>
>>> I couldn't find the code anywhere that enforces word-aligned accesses to
>>> these registers.  Do we have that?
>>
>> Not that I am aware of. I was suggesting this since we have one in the
>> IROUTER function. Architecturally we don't need to support halfword
>> accesses, it's: byte + word, word only or double-word + word, depending
>> on the actual register, IIRC.
>> As a fix we can at least deny (read: ignore) halfword accesses in
>> general in the dispatcher. Shall I do this (two two-liners)?
>> I think byte and word accesses are safe with the existing handlers last
>> time I checked.
>>
>>> If that's not the case, doesn't this break of you do a non-word aligned
>>> access?
>>
>> Why would it? vgic_data_host_to_mmio_bus and extract_bytes should cover
>> this, shouldn't they?
>>
> 
> I think this breaks on a simple byte access.  Let's say you are
> accessing byte 1 (addr & 0x7ff == 1), then because you start your loop
> at 0, you're going to set bits [7:0] in the value variable, and then
> extract bits [15:8] in extract_bytes(), right?

Looks like, yes. I thought that it would take care of that, because a
handler function shouldn't be bothered with this: it just acts on the
lower bytes (till the length) and extract_bytes() does the rest. I think
this is what the old magic vgic_reg_access does due to "regval >>
word_offset".

Do you care to fix this in extract_bytes?

Cheers,
Andre.