linux-kernel - Re: [PATCH 16/16] KVM: arm64: pkvm: Unshare guest structs during teardown

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YW6SiMLU19YN1ARe@google.com>
Date:   Tue, 19 Oct 2021 10:40:24 +0100
From:   Quentin Perret <qperret@...gle.com>
To:     Marc Zyngier <maz@...nel.org>
Cc:     James Morse <james.morse@....com>,
        Alexandru Elisei <alexandru.elisei@....com>,
        Suzuki K Poulose <suzuki.poulose@....com>,
        Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will@...nel.org>, Fuad Tabba <tabba@...gle.com>,
        David Brazdil <dbrazdil@...gle.com>,
        linux-arm-kernel@...ts.infradead.org, kvmarm@...ts.cs.columbia.edu,
        linux-kernel@...r.kernel.org, kernel-team@...roid.com
Subject: Re: [PATCH 16/16] KVM: arm64: pkvm: Unshare guest structs during
 teardown

On Monday 18 Oct 2021 at 18:12:22 (+0100), Marc Zyngier wrote:
> On 2021-10-18 15:03, Quentin Perret wrote:
> > On Monday 18 Oct 2021 at 11:32:13 (+0100), Quentin Perret wrote:
> > > Another option is to take a refcount on 'current' from
> > > kvm_arch_vcpu_run_map_fp() before sharing thread-specific structs with
> > > the hyp and release the refcount of the previous task after unsharing.
> > > But that means we'll have to also drop the refcount when the vcpu
> > > gets destroyed, as well as explicitly unshare at that point. Shouldn't
> > > be too bad I think. Thoughts?
> > 
> > Something like the below seems to work OK on my setup, including
> > SIGKILL'ing the guest and such. How much do you hate it?
> 
> It is annoyingly elegant! Small nitpick below.
> 
> > 
> > diff --git a/arch/arm64/include/asm/kvm_host.h
> > b/arch/arm64/include/asm/kvm_host.h
> > index f8be56d5342b..50598d704c71 100644
> > --- a/arch/arm64/include/asm/kvm_host.h
> > +++ b/arch/arm64/include/asm/kvm_host.h
> > @@ -322,6 +322,7 @@ struct kvm_vcpu_arch {
> > 
> >  	struct thread_info *host_thread_info;	/* hyp VA */
> >  	struct user_fpsimd_state *host_fpsimd_state;	/* hyp VA */
> > +	struct task_struct *parent_task;
> > 
> >  	struct {
> >  		/* {Break,watch}point registers */
> > @@ -738,6 +739,7 @@ int kvm_arch_vcpu_run_map_fp(struct kvm_vcpu *vcpu);
> >  void kvm_arch_vcpu_load_fp(struct kvm_vcpu *vcpu);
> >  void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu);
> >  void kvm_arch_vcpu_put_fp(struct kvm_vcpu *vcpu);
> > +void kvm_vcpu_unshare_task_fp(struct kvm_vcpu *vcpu);
> > 
> >  static inline bool kvm_pmu_counter_deferred(struct perf_event_attr
> > *attr)
> >  {
> > diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c
> > index 2fe1128d9f3d..27afeebbe1cb 100644
> > --- a/arch/arm64/kvm/fpsimd.c
> > +++ b/arch/arm64/kvm/fpsimd.c
> > @@ -15,6 +15,22 @@
> >  #include <asm/kvm_mmu.h>
> >  #include <asm/sysreg.h>
> > 
> > +void kvm_vcpu_unshare_task_fp(struct kvm_vcpu *vcpu)
> > +{
> > +	struct task_struct *p = vcpu->arch.parent_task;
> > +	struct user_fpsimd_state *fpsimd;
> > +	struct thread_info *ti;
> > +
> > +	if (!static_branch_likely(&kvm_protected_mode_initialized) || !p)
> 
> Shouldn't this be a check on is_protected_kvm_enabled() instead?
> The two should be equivalent outside of the initialisation code...

Yup, it'd be nice to do checks on kvm_protected_mode_initialized only
when they're strictly necessary, and that's not the case here. I'll fold
that change in v2.

Cheers
Quentin