[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20150508164649.GK24744@cbox>
Date: Fri, 8 May 2015 18:46:49 +0200
From: Christoffer Dall <christoffer.dall@...aro.org>
To: Alex Bennée <alex.bennee@...aro.org>
Cc: kvm@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
kvmarm@...ts.cs.columbia.edu, marc.zyngier@....com,
peter.maydell@...aro.org, agraf@...e.de, drjones@...hat.com,
pbonzini@...hat.com, zhichao.huang@...aro.org,
jan.kiszka@...mens.com, dahi@...ux.vnet.ibm.com,
r65777@...escale.com, bp@...e.de, Gleb Natapov <gleb@...nel.org>,
Catalin Marinas <catalin.marinas@....com>,
Will Deacon <will.deacon@....com>,
open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v3 10/12] KVM: arm64: trap nested debug register access
On Thu, May 07, 2015 at 10:07:13AM +0100, Alex Bennée wrote:
> When we are using the hardware registers for guest debug we need to deal
> with the guests access to them. There is already a mechanism for dealing
> with these accesses so we build on top of that.
>
> - any access to mdscr_el1 is now stored in the mirror location
> - access to DBG[WB][CV]R continues to go to guest's context
>
> There is one register (MDCCINT_EL1) which guest debug doesn't care about
> so this behaves as before.
>
> Signed-off-by: Alex Bennée <alex.bennee@...aro.org>
>
> ---
> v3
> - re-factor for better flow and fall through.
> - much simpler with debug_ptr (use the guest area as before)
> - tweak shadow fn to avoid multi-line if
>
> diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
> index a44fb32..7aa3b3a 100644
> --- a/arch/arm64/include/asm/kvm_host.h
> +++ b/arch/arm64/include/asm/kvm_host.h
> @@ -132,7 +132,13 @@ struct kvm_vcpu_arch {
> * here.
> */
>
> - /* Guest registers we preserve during guest debugging */
> + /*
> + * Guest registers we preserve during guest debugging.
> + *
> + * These shadow registers are updated by the kvm_handle_sys_reg
> + * trap handler if the guest accesses or updates them while we
> + * are using guest debug.
> + */
> struct {
> u32 pstate;
> u32 mdscr_el1;
> diff --git a/arch/arm64/kvm/debug.c b/arch/arm64/kvm/debug.c
> index 1ab63dd..dc8bca8 100644
> --- a/arch/arm64/kvm/debug.c
> +++ b/arch/arm64/kvm/debug.c
> @@ -50,8 +50,7 @@ static void restore_guest_debug_regs(struct kvm_vcpu *vcpu)
> {
> *vcpu_cpsr(vcpu) |=
> (vcpu->arch.guest_debug_state.pstate & SPSR_DEBUG_MASK);
> - vcpu_sys_reg(vcpu, MDSCR_EL1) |=
> - (vcpu->arch.guest_debug_state.mdscr_el1 & MDSCR_EL1_DEBUG_MASK);
> + vcpu_sys_reg(vcpu, MDSCR_EL1) = vcpu->arch.guest_debug_state.mdscr_el1;
> }
>
> /**
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index c370b40..95f422f 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -196,11 +196,40 @@ static bool trap_dbgauthstatus_el1(struct kvm_vcpu *vcpu,
> * - If the dirty bit is set, save guest registers, restore host
> * registers and clear the dirty bit. This ensure that the host can
> * now use the debug registers.
> + *
> + * We also use this mechanism to set-up the debug registers for guest
s/set-up/set up/
> + * debugging. If this is the case we want to ensure the guest sees
If this is the case, (comma)
> + * the right versions of the registers - even if they are not going to
> + * be effective while guest debug is using HW debug.
> + *
> */
> +
> +static bool shadow_debug_reg(struct kvm_vcpu *vcpu,
> + const struct sys_reg_params *p,
> + const struct sys_reg_desc *r)
> +{
> + /* MDSCR_EL1 */
> + if (r->reg == MDSCR_EL1) {
> + u32 *shadow_mdscr_el1 = &vcpu->arch.guest_debug_state.mdscr_el1;
> +
> + if (p->is_write)
> + *shadow_mdscr_el1 = *vcpu_reg(vcpu, p->Rt);
> + else
> + *vcpu_reg(vcpu, p->Rt) = *shadow_mdscr_el1;
> +
> + return true;
> + }
> +
> + return false;
> +}
> +
> static bool trap_debug_regs(struct kvm_vcpu *vcpu,
> const struct sys_reg_params *p,
> const struct sys_reg_desc *r)
> {
> + if (vcpu->guest_debug && shadow_debug_reg(vcpu, p, r))
> + return true;
> +
so do we also have a MDSCR_EL1 in sys_regs and one in guest_debug_state
now?
If yes, what are the differences between the two?
> if (p->is_write) {
> vcpu_sys_reg(vcpu, r->reg) = *vcpu_reg(vcpu, p->Rt);
> vcpu->arch.debug_flags |= KVM_ARM64_DEBUG_DIRTY;
> --
> 2.3.5
>
Thanks,
-Christoffer
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists