From: marc.zyngier@arm•com (Marc Zyngier)
To: linux-arm-kernel@lists•infradead.org
Subject: [PATCH v3 17/19] arm64: KVM: move hcr_el2 setting into vgic-v2-switch.S
Date: Wed, 14 May 2014 17:58:04 +0100 [thread overview]
Message-ID: <5373A09C.9020306@arm.com> (raw)
In-Reply-To: <CAMJs5B9RwaeZhxoWnz1FzmX9swsA32Gqz9NwLZSQiFuGJ-v5Lg@mail.gmail.com>
On 14/05/14 17:34, Christoffer Dall wrote:
> On 14 May 2014 15:33, Marc Zyngier <marc.zyngier@arm•com> wrote:
>> On Fri, May 09 2014 at 3:07:23 pm BST, Christoffer Dall <christoffer.dall@linaro•org> wrote:
>>> On Wed, Apr 16, 2014 at 02:39:49PM +0100, Marc Zyngier wrote:
>>>> GICv3 requires the hcr_el2 switch to be tightly coupled with some
>>>> of the interrupt controller's register switch.
>>>
>>> can you be more specific, this feels a bit odd, enabling Stage-2
>>> translation and configuring all traps from within the vgic code...
>>
>> The IMO and FMO bits must be set before restoring the various system
>> registers in GICv3. But I agreee that this looks pretty horrible.
>>
>> The alternative is to split the bits we set in HCR_EL2 into two sets (VM
>> and trap control on one side, interrupt control on the other). This
>> would translate into two accesses to HCR_EL2, but it would look
>> nicer. I'll have a look.
>>
>>>> In order to have similar code paths, start moving the hcr_el2
>>>> manipulation code to the GICv2 switch code.
>>>>
>>>> Acked-by: Catalin Marinas <catalin.marinas@arm•com>
>>>> Signed-off-by: Marc Zyngier <marc.zyngier@arm•com>
>>>> ---
>>>> arch/arm64/kvm/hyp.S | 7 -------
>>>> arch/arm64/kvm/vgic-v2-switch.S | 8 ++++++++
>>>> 2 files changed, 8 insertions(+), 7 deletions(-)
>>>>
>>>> diff --git a/arch/arm64/kvm/hyp.S b/arch/arm64/kvm/hyp.S
>>>> index aed72d0..92b9120 100644
>>>> --- a/arch/arm64/kvm/hyp.S
>>>> +++ b/arch/arm64/kvm/hyp.S
>>>> @@ -335,11 +335,6 @@
>>>> .endm
>>>>
>>>> .macro activate_traps
>>>> - ldr x2, [x0, #VCPU_IRQ_LINES]
>>>> - ldr x1, [x0, #VCPU_HCR_EL2]
>>>> - orr x2, x2, x1
>>>> - msr hcr_el2, x2
>>>> -
>>>> ldr x2, =(CPTR_EL2_TTA)
>>>> msr cptr_el2, x2
>>>>
>>>> @@ -353,8 +348,6 @@
>>>> .endm
>>>>
>>>> .macro deactivate_traps
>>>> - mov x2, #HCR_RW
>>>> - msr hcr_el2, x2
>>>> msr cptr_el2, xzr
>>>> msr hstr_el2, xzr
>>>>
>>>> diff --git a/arch/arm64/kvm/vgic-v2-switch.S b/arch/arm64/kvm/vgic-v2-switch.S
>>>> index c5dc777..d36cd7a 100644
>>>> --- a/arch/arm64/kvm/vgic-v2-switch.S
>>>> +++ b/arch/arm64/kvm/vgic-v2-switch.S
>>>> @@ -85,6 +85,9 @@ CPU_BE( rev w5, w5 )
>>>> sub w4, w4, #1
>>>> cbnz w4, 1b
>>>> 2:
>>>> + mov x2, #HCR_RW
>>>> + msr hcr_el2, x2
>>>> + isb
>>>> .endm
>>>>
>>>> /*
>>>> @@ -92,6 +95,11 @@ CPU_BE( rev w5, w5 )
>>>> * x0: Register pointing to VCPU struct
>>>> */
>>>> .macro restore_vgic_v2_state
>>>> + ldr x2, [x0, #VCPU_IRQ_LINES]
>>>
>>> will this ever have any values on aarch64? Don't we mandate vgic
>>> support and bail out during hyp init if we cannot init a vgic?
>>
>> Yes. But that doesn't mean we don't support the feature either. The case
>> is fairly slim, I agree, but it has been there since Day-1...
>>
> See kvm_vm_ioctl_irq_line() in arch/arm/kvm/arm.c:
>
> case KVM_ARM_IRQ_TYPE_CPU:
> if (irqchip_in_kernel(kvm))
> return -ENXIO;
Unfortunately, this only checks if the VM has a vgic instantiated. It is
always possible to create a VM without the in-kernel GIC, and use the
pins to inject IRQs. As I said, unlikely to happen, but nonetheless...
M.
--
Jazz is not dead. It just smells funny...
next prev parent reply other threads:[~2014-05-14 16:58 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-04-16 13:39 [PATCH v3 00/19] arm64: GICv3 support Marc Zyngier
2014-04-16 13:39 ` [PATCH v3 01/19] ARM: GIC: move some bits of GICv2 to a library-type file Marc Zyngier
2014-05-09 14:05 ` Christoffer Dall
2014-05-12 16:29 ` Marc Zyngier
2014-04-16 13:39 ` [PATCH v3 02/19] arm64: initial support for GICv3 Marc Zyngier
2014-05-09 14:05 ` Christoffer Dall
2014-05-12 16:54 ` Marc Zyngier
2014-05-14 16:02 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 03/19] arm64: GICv3 device tree binding documentation Marc Zyngier
2014-05-09 14:05 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 04/19] arm64: boot protocol documentation update for GICv3 Marc Zyngier
2014-05-09 14:05 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 05/19] KVM: arm/arm64: vgic: move GICv2 registers to their own structure Marc Zyngier
2014-05-09 14:05 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 06/19] KVM: ARM: vgic: introduce vgic_ops and LR manipulation primitives Marc Zyngier
2014-05-09 14:05 ` Christoffer Dall
2014-05-12 17:28 ` Marc Zyngier
2014-05-14 16:17 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 07/19] KVM: ARM: vgic: abstract access to the ELRSR bitmap Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-05-12 17:41 ` Marc Zyngier
2014-04-16 13:39 ` [PATCH v3 08/19] KVM: ARM: vgic: abstract EISR bitmap access Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 09/19] KVM: ARM: vgic: abstract MISR decoding Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 10/19] KVM: ARM: vgic: move underflow handling to vgic_ops Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 11/19] KVM: ARM: vgic: abstract VMCR access Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-05-13 17:43 ` Marc Zyngier
2014-05-14 16:28 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 12/19] KVM: ARM: vgic: introduce vgic_enable Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 13/19] KVM: ARM: introduce vgic_params structure Marc Zyngier
2014-05-09 14:06 ` Christoffer Dall
2014-05-12 17:50 ` Marc Zyngier
2014-04-16 13:39 ` [PATCH v3 14/19] KVM: ARM: vgic: split GICv2 backend from the main vgic code Marc Zyngier
2014-05-09 14:07 ` Christoffer Dall
2014-05-12 17:54 ` Marc Zyngier
2014-04-16 13:39 ` [PATCH v3 15/19] arm64: KVM: remove __kvm_hyp_code_{start, end} from hyp.S Marc Zyngier
2014-05-09 14:07 ` [PATCH v3 15/19] arm64: KVM: remove __kvm_hyp_code_{start,end} " Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 16/19] arm64: KVM: split GICv2 world switch from hyp code Marc Zyngier
2014-05-09 14:07 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 17/19] arm64: KVM: move hcr_el2 setting into vgic-v2-switch.S Marc Zyngier
2014-05-09 14:07 ` Christoffer Dall
2014-05-14 14:33 ` Marc Zyngier
2014-05-14 16:34 ` Christoffer Dall
2014-05-14 16:58 ` Marc Zyngier [this message]
2014-05-15 12:20 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 18/19] KVM: ARM: vgic: add the GICv3 backend Marc Zyngier
2014-05-09 14:07 ` Christoffer Dall
2014-05-14 17:47 ` Marc Zyngier
2014-05-15 8:13 ` Marc Zyngier
2014-05-15 12:18 ` Christoffer Dall
2014-04-16 13:39 ` [PATCH v3 19/19] arm64: KVM: vgic: add GICv3 world switch Marc Zyngier
2014-05-09 14:07 ` Christoffer Dall
2014-05-15 8:31 ` Marc Zyngier
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5373A09C.9020306@arm.com \
--to=marc.zyngier@arm$(echo .)com \
--cc=linux-arm-kernel@lists$(echo .)infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox