2018-02-08 11:59:12

by Jérémy Fanguède

[permalink] [raw]
Subject: [PATCH] KVM: arm64: Enable the EL1 physical timer for AArch32 guests

Some 32bits guest OS can use the CNTP timer, however KVM does not
handle the accesses, injecting a fault instead.

Use the proper handlers to emulate the EL1 Physical Timer (CNTP)
register accesses of AArch32 guests.

Signed-off-by: Jérémy Fanguède <[email protected]>
Signed-off-by: Alvise Rigo <[email protected]>
---
arch/arm64/kvm/sys_regs.c | 6 ++++++
1 file changed, 6 insertions(+)

diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 50a43c7..c0ab4f7 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -1545,6 +1545,11 @@ static const struct sys_reg_desc cp15_regs[] = {

{ Op1( 0), CRn(13), CRm( 0), Op2( 1), access_vm_reg, NULL, c13_CID },

+ /* CNTP_TVAL */
+ { Op1( 0), CRn(14), CRm( 2), Op2( 0), access_cntp_tval },
+ /* CNTP_CTL */
+ { Op1( 0), CRn(14), CRm( 2), Op2( 1), access_cntp_ctl },
+
/* PMEVCNTRn */
PMU_PMEVCNTR(0),
PMU_PMEVCNTR(1),
@@ -1618,6 +1623,7 @@ static const struct sys_reg_desc cp15_64_regs[] = {
{ Op1( 0), CRn( 0), CRm( 9), Op2( 0), access_pmu_evcntr },
{ Op1( 0), CRn( 0), CRm(12), Op2( 0), access_gic_sgi },
{ Op1( 1), CRn( 0), CRm( 2), Op2( 0), access_vm_reg, NULL, c2_TTBR1 },
+ { Op1( 2), CRn( 0), CRm(14), Op2( 0), access_cntp_cval },
};

/* Target specific emulation tables */
--
2.7.4



2018-02-08 12:09:39

by Marc Zyngier

[permalink] [raw]
Subject: Re: [PATCH] KVM: arm64: Enable the EL1 physical timer for AArch32 guests

On 08/02/18 11:57, Jérémy Fanguède wrote:
> Some 32bits guest OS can use the CNTP timer, however KVM does not
> handle the accesses, injecting a fault instead.
>
> Use the proper handlers to emulate the EL1 Physical Timer (CNTP)
> register accesses of AArch32 guests.
>
> Signed-off-by: Jérémy Fanguède <[email protected]>
> Signed-off-by: Alvise Rigo <[email protected]>
> ---
> arch/arm64/kvm/sys_regs.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index 50a43c7..c0ab4f7 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -1545,6 +1545,11 @@ static const struct sys_reg_desc cp15_regs[] = {
>
> { Op1( 0), CRn(13), CRm( 0), Op2( 1), access_vm_reg, NULL, c13_CID },
>
> + /* CNTP_TVAL */
> + { Op1( 0), CRn(14), CRm( 2), Op2( 0), access_cntp_tval },
> + /* CNTP_CTL */
> + { Op1( 0), CRn(14), CRm( 2), Op2( 1), access_cntp_ctl },
> +
> /* PMEVCNTRn */
> PMU_PMEVCNTR(0),
> PMU_PMEVCNTR(1),
> @@ -1618,6 +1623,7 @@ static const struct sys_reg_desc cp15_64_regs[] = {
> { Op1( 0), CRn( 0), CRm( 9), Op2( 0), access_pmu_evcntr },
> { Op1( 0), CRn( 0), CRm(12), Op2( 0), access_gic_sgi },
> { Op1( 1), CRn( 0), CRm( 2), Op2( 0), access_vm_reg, NULL, c2_TTBR1 },
> + { Op1( 2), CRn( 0), CRm(14), Op2( 0), access_cntp_cval },
> };
>
> /* Target specific emulation tables */
>

Seems OK to me. Can you please update the corresponding 32bit code while
you're at it?

Thanks,

M.
--
Jazz is not dead. It just smells funny...