From: Marc Zyngier <maz@kernel.org>
To: Oliver Upton <oupton@google.com>
Cc: kvmarm@lists.cs.columbia.edu, kvm@vger.kernel.org,
James Morse <james.morse@arm.com>,
Alexandru Elisei <alexandru.elisei@arm.com>,
Suzuki K Poulose <suzuki.poulose@arm.com>,
linux-arm-kernel@lists.infradead.org,
Andrew Jones <drjones@redhat.com>,
Peter Shier <pshier@google.com>,
Ricardo Koller <ricarkol@google.com>,
Reiji Watanabe <reijiw@google.com>
Subject: Re: [PATCH v3 4/6] KVM: arm64: Emulate the OS Lock
Date: Mon, 29 Nov 2021 14:15:46 +0000 [thread overview]
Message-ID: <87r1azm4j1.wl-maz@kernel.org> (raw)
In-Reply-To: <20211123210109.1605642-5-oupton@google.com>
On Tue, 23 Nov 2021 21:01:07 +0000,
Oliver Upton <oupton@google.com> wrote:
>
> The OS lock blocks all debug exceptions at every EL. To date, KVM has
> not implemented the OS lock for its guests, despite the fact that it is
> mandatory per the architecture. Simple context switching between the
> guest and host is not appropriate, as its effects are not constrained to
> the guest context.
>
> Emulate the OS Lock by clearing MDE and SS in MDSCR_EL1, thereby
> blocking all but software breakpoint instructions. To handle breakpoint
> instructions, trap debug exceptions to EL2 and skip the instruction.
Skipping breakpoint instructions? I don't think you can do that, as
the guest does rely on BRK always being effective. I also don't see
where you do that...
>
> Signed-off-by: Oliver Upton <oupton@google.com>
> ---
> arch/arm64/include/asm/kvm_host.h | 4 ++++
> arch/arm64/kvm/debug.c | 27 +++++++++++++++++++++++----
> arch/arm64/kvm/sys_regs.c | 6 +++---
> 3 files changed, 30 insertions(+), 7 deletions(-)
>
> diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
> index 53fc8a6eaf1c..e5a06ff1cba6 100644
> --- a/arch/arm64/include/asm/kvm_host.h
> +++ b/arch/arm64/include/asm/kvm_host.h
> @@ -726,6 +726,10 @@ void kvm_arm_vcpu_init_debug(struct kvm_vcpu *vcpu);
> void kvm_arm_setup_debug(struct kvm_vcpu *vcpu);
> void kvm_arm_clear_debug(struct kvm_vcpu *vcpu);
> void kvm_arm_reset_debug_ptr(struct kvm_vcpu *vcpu);
> +
> +#define kvm_vcpu_os_lock_enabled(vcpu) \
> + (!!(__vcpu_sys_reg(vcpu, OSLSR_EL1) & SYS_OSLSR_OSLK))
> +
> int kvm_arm_vcpu_arch_set_attr(struct kvm_vcpu *vcpu,
> struct kvm_device_attr *attr);
> int kvm_arm_vcpu_arch_get_attr(struct kvm_vcpu *vcpu,
> diff --git a/arch/arm64/kvm/debug.c b/arch/arm64/kvm/debug.c
> index db9361338b2a..7835c76347ce 100644
> --- a/arch/arm64/kvm/debug.c
> +++ b/arch/arm64/kvm/debug.c
> @@ -53,6 +53,14 @@ static void restore_guest_debug_regs(struct kvm_vcpu *vcpu)
> vcpu_read_sys_reg(vcpu, MDSCR_EL1));
> }
>
> +/*
> + * Returns true if the host needs to use the debug registers.
> + */
> +static inline bool host_using_debug_regs(struct kvm_vcpu *vcpu)
> +{
> + return vcpu->guest_debug || kvm_vcpu_os_lock_enabled(vcpu);
Just the name of the function has sent my head spinning. Even if the
*effects* of the host debug and the OS Lock are vaguely similar from
the guest PoV, they really are different things, and I'd rather not
lob them together.
> +}
> +
> /**
> * kvm_arm_init_debug - grab what we need for debug
> *
> @@ -105,9 +113,11 @@ static void kvm_arm_setup_mdcr_el2(struct kvm_vcpu *vcpu)
> * - Userspace is using the hardware to debug the guest
> * (KVM_GUESTDBG_USE_HW is set).
> * - The guest is not using debug (KVM_ARM64_DEBUG_DIRTY is clear).
> + * - The guest has enabled the OS Lock (debug exceptions are blocked).
> */
> if ((vcpu->guest_debug & KVM_GUESTDBG_USE_HW) ||
> - !(vcpu->arch.flags & KVM_ARM64_DEBUG_DIRTY))
> + !(vcpu->arch.flags & KVM_ARM64_DEBUG_DIRTY) ||
> + kvm_vcpu_os_lock_enabled(vcpu))
> vcpu->arch.mdcr_el2 |= MDCR_EL2_TDA;
>
> trace_kvm_arm_set_dreg32("MDCR_EL2", vcpu->arch.mdcr_el2);
> @@ -160,8 +170,10 @@ void kvm_arm_setup_debug(struct kvm_vcpu *vcpu)
>
> kvm_arm_setup_mdcr_el2(vcpu);
>
> - /* Is Guest debugging in effect? */
> - if (vcpu->guest_debug) {
> + /*
> + * Check if we need to use the debug registers.
> + */
> + if (host_using_debug_regs(vcpu)) {
I'd rather you expand the helper here and add the comment you have in
the commit message explaining the machine-wide effect of the OS Lock.
> /* Save guest debug state */
> save_guest_debug_regs(vcpu);
>
> @@ -223,6 +235,10 @@ void kvm_arm_setup_debug(struct kvm_vcpu *vcpu)
> trace_kvm_arm_set_regset("WAPTS", get_num_wrps(),
> &vcpu->arch.debug_ptr->dbg_wcr[0],
> &vcpu->arch.debug_ptr->dbg_wvr[0]);
> + } else if (kvm_vcpu_os_lock_enabled(vcpu)) {
> + mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
> + mdscr &= ~DBG_MDSCR_MDE;
> + vcpu_write_sys_reg(vcpu, mdscr, MDSCR_EL1);
> }
> }
>
> @@ -244,7 +260,10 @@ void kvm_arm_clear_debug(struct kvm_vcpu *vcpu)
> {
> trace_kvm_arm_clear_debug(vcpu->guest_debug);
>
> - if (vcpu->guest_debug) {
> + /*
> + * Restore the guest's debug registers if we were using them.
> + */
> + if (host_using_debug_regs(vcpu)) {
> restore_guest_debug_regs(vcpu);
>
> /*
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index 5dbdb45d6d44..1346906f5c46 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -1453,9 +1453,9 @@ static unsigned int mte_visibility(const struct kvm_vcpu *vcpu,
> * Debug handling: We do trap most, if not all debug related system
> * registers. The implementation is good enough to ensure that a guest
> * can use these with minimal performance degradation. The drawback is
> - * that we don't implement any of the external debug, none of the
> - * OSlock protocol. This should be revisited if we ever encounter a
> - * more demanding guest...
> + * that we don't implement any of the external debug architecture.
> + * This should be revisited if we ever encounter a more demanding
> + * guest...
> */
> static const struct sys_reg_desc sys_reg_descs[] = {
> { SYS_DESC(SYS_DC_ISW), access_dcsw },
Thanks,
M.
--
Without deviation from the norm, progress is not possible.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2021-11-29 14:17 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-11-23 21:01 [PATCH v3 0/6] KVM: arm64: Emulate the OS lock Oliver Upton
2021-11-23 21:01 ` [PATCH v3 1/6] KVM: arm64: Correctly treat writes to OSLSR_EL1 as undefined Oliver Upton
2021-11-23 21:01 ` [PATCH v3 2/6] KVM: arm64: Stash OSLSR_EL1 in the cpu context Oliver Upton
2021-11-23 21:01 ` [PATCH v3 3/6] KVM: arm64: Allow guest to set the OSLK bit Oliver Upton
2021-11-29 11:50 ` Marc Zyngier
2021-12-06 17:39 ` Oliver Upton
2021-12-06 18:47 ` Marc Zyngier
2021-11-23 21:01 ` [PATCH v3 4/6] KVM: arm64: Emulate the OS Lock Oliver Upton
2021-11-29 14:15 ` Marc Zyngier [this message]
2021-12-06 17:34 ` Oliver Upton
2021-11-23 21:01 ` [PATCH v3 5/6] selftests: KVM: Add OSLSR_EL1 to the list of blessed regs Oliver Upton
2021-11-23 21:01 ` [PATCH v3 6/6] selftests: KVM: Test OS lock behavior Oliver Upton
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=87r1azm4j1.wl-maz@kernel.org \
--to=maz@kernel.org \
--cc=alexandru.elisei@arm.com \
--cc=drjones@redhat.com \
--cc=james.morse@arm.com \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.cs.columbia.edu \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=oupton@google.com \
--cc=pshier@google.com \
--cc=reijiw@google.com \
--cc=ricarkol@google.com \
--cc=suzuki.poulose@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
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;
as well as URLs for NNTP newsgroup(s).